The HEDT Haswell chip had quad channel memory that gives the cpu access to a theoretical maximum memory bandwidth of about 85GB/s less a bit of over head. The Ringbus at 4.2ghz creates a pipe that connects the cpu cores to the installed memory and to the PCIe controller through a pipe that can potentially carry 134.4GB/s of data traffic. the 12100 Ringbus, if it runds at the same 4.2Ghz, creates a pipe of similar capacity to the 5950X but with the 3200 Ram can only get memory data at a maximum of 51.2GB/s less a bit of overhead. neither cpus can saturate the ring and there is a lot of extra capacity available for GPU data traffic that the CPU is generating while playing a game On the Zen 1 and Zen 2 CPUs the infinity fabric performs the same function as the ring bus. The frequency is set to match the memory clock. if you have 3200 Ram installed, you have the Infinity Fabric running at 1600Mhz. The Ram can be delivered to the CPU at a maximum of 51.2GB/s just like the 12100. The infinity Fabric interconnects which are clocked the same as memory creates pipe that can carry 51.2GB/s. there is no extra overhead available if the ram is being read at max capacity to transfer data between CPU and GPU. Ram read capacity has to be compromised to allow the traffic to the GPU to share the same interconnect. If you want to do something educational, try setting the Ringbus in the two intel boxes to a frequency that exactly matches the infinity fabric FCLK on the AMD machines. The memory frequency on the HEDT platform doesn't matter cause u have quad channel memory installed an an abundance of memory that is limited to 51.2GB/s anyway. All of the dual channel systems should be able to run 3200 DDR4 as the lowest common denominator so select that and then set the ringbus to 1.6GHz so it matches the IF pipes. I think that you may find the changes in performance on the intel systems, compared to the Ryzen systems interesting. I would also suggest running cinebench on each system. It mostly runs out of the cache so it is a good measure for creating a baseline of the relative ipc of each CPU without being too much by the memory speed. You should see the performance variations in games look similar to the variations in the cinebench scores rather than what we are all used to seeing. The test will show how hobbled Ryzen was with zen1 and how it improved over the generations and also demonstrate why Ryzen traditionally struggled with gaming workloads before and not now with the 5800x. The 5800x has double the infinity fabric interconnect capacity so it behaves more like an intel cpu but It can also be used to show that the main improvement in performance in Ryzen comes primarily from the bigger pipe created by the faster FCLK, Faster Ram certainly provides a bit of the boost to performance but it is not, as everyone seems to assume, from latency and brute strength of the faster silicon in the 3800/4000 ddr4 chips. Clearly this is a test for science, not a recommendation to daily your rig with ringbus set well below the default settings.
@@informedchoice2249 In conceptual terms, a Ring bus is like a 4 lane ringroad highway that connects the cores to the memory and the gpu/pcie bus with on/off ramps at each location. The data can go in either direction to get to the other destination without congestion. Infinity fabric is built like a two lane road system with an intersection in the middle. the io die acts like the traffic lights managing the traffic. there is only the one path between each destination going through the central crossroads. Like in the roads, the traffic needs to be managed to prevent collisions at the intersecton which leads to some traffic congestion at periods of heavy traffic
@@yasirrakhurrafat1142 I know hot to utilize even server grade motherboards to their fullest potential, that doesn't mean that if I was building a gaming PC be first or second hand I'd care for them. We are talking pure gaming here. I'd rather get a ryzen 5 3600 with a B450/550 board second hand for a total of 200-250$ than a I7 5960x for 80$(here it's 110-130$) and a X99 motherboard for close to 250-350$. Also the E5 2620v3 and 2630 V3 go for like 20-40$, but they can't be overclocked as easily and as much.
@@badass6300 Irrelevant but , isn't the x99 socket setup considered HEDT ? And yeah you're right , low cpu intensive titles run well on max on even 12100 . While providing high performance . I was just trying to engage you to the endless possibilities of the x99/c602 chipset , for the budget . Just imagine getting the 2699v3 oc it to the max , and using it for multiple gaming sessions locally or remote , through VMs or Multiseat . Rather than building 8 boxes , you'd build have to build 2 , for medium spec Gaming on each instance .
@@laughingcross8108 its kind of a individual preference , as to how much performance you want per vm . But if you're gaming , two users/vms , top demanding games , you should be good with an i7 12th gen or a ryzen 5800 , with 32gb of ram and 3050s( per vm ) or 3060s( per vm ) . For absolute above 60 fps performance in 1080p high in all games , 1440p high 60 fps in some .
I wasn’t surprised to see possibly that the Intel 5960x would at least come close to Zen 2 ….. but Zen 3 was a mind blown 🤯🤯🤯 situation for me! Absolutely lovin 💕 this series of tweaking memory timings on different generations of AMD / Intel CPUs’. Truely eye opening a number of these results 🫠🤩🤩
Back in the day I had a rig with the 5960X with a pair of GTX 980Ti's and it was a beast for it's time. The 5960X is still a beast of a CPU even against some of the newer stuff.
If this isn't a full fledged endorsement of Overclocking then I don't know what is! These results are mind-blowing and gives me hope that people in the future will learn how important properly tuning your system can be. Great video!
I recently upgraded my i7 5820k system to a 5960x, overclocked. I think I'll be able to get another few years out of this machine. That will make it the first computer I get a full decade out of without upgrading the motherboard. Intel really outdid themselves on this CPU for the time.
Same here my friend. Rocking the ASUS Rampage V E10 mobo with 1000w H1000x psu and 3200 Corsair RGB RAM with a FE 3090 . I run a 5k screen and maxing out 150-200% MSAA or Full ray tracing without DLSS garbage, I rock 60fps all day and that is all I will ever ask for.
@Vinson Smith I mainly use it for video games and it isn't that much different. The 5960x was way over spec for what I needed but I was keen then and I liked to have the best.
le 8700K est bien meilleur que ce I7 5960X ... il ne faut pas oublier qu'il faut OC à l'extrême ce cpu, et que ça engendre une consommation de plus de 300W
It's 8 cores/16 threads. I have a 9900K and it only has 16 MB of L3 cache, while the 5960X has 20 MB. The size of the L3 cache really helps with gaming performance.
Im not amazed for a simple reason that I was using i7-3930K @4500Mhz + DDR3 4x4 2400 11-13-12-20-1 for about 9 years, until I bought i9-12900K because.... reasons... I guess, but the 3930K didn't gave me any doubts that I need a new platform, and it was so stable and strong still. You need to check it next. ! :) It is a legendary platform (X79). Good work dude ! ;)
There was a time a few years ago when one could get top end X79 mbds very cheap, I bought a bunch of them, mainly the ASUS P9X79/E WS. Best deal I bought was an ASUS Rampage IV Extreme, 3930K C2 and 16GB/2400 RAM for just 200 UKP (it formed the basis of a gaming PC I built for a friend, though the CPU was swapped out for a 4930K).
I moved from my x99 5960x to x299 and 7900x, more cores, a bit higher frequencies, but shame there's lesser l3 cache though on skylake-x chip.. Looking to get 9900x or even 10900x in time the price drops even more. Going to delid my 7900x in the mean time.. It's awesome these chips are now so affordable, I'm not even looking at current platforms, my HEDT is still going strong and will be good for another 5+ years easy.
There problem back then was that the Core i7 5960X was a $999 price and the motherboards was expensive too. You have to overclock it to really compete, which will reduce the live span of an already old CPU released on 2014. On today prices is a great project to run until it die, but was not possible at the R7 5800 release date, since the Core i7 5960X used was at ~$500-600 price on a expensive platform. The i3 12100 still in my opinion the best pure gaming machine for mid-high gaming target. But what I want too see is benchmarks where the machine have like a chrome with 20+ taps, two monitors, playing twich or youtube and other background application running at the same time you game, there is where Ryzen shine. My use case will never be just the game running.
@@lmaoitsmee1812 if your talking about live span when overclock....it always the motherboard that will be stressed more than the cpu. Between the two, I would say the motherboard wouldn't even hold up unless it's a really high end one. The cpu will always survive. Rarely dead, really rare. If you suspect the cpu is dead, it always the motherboard that is the problem. X99 is kinda still expensive, I would rather go for 5600X regardless, newer and easy to get 😂😂 12100 is the best bang for the buck, period!
@@lmaoitsmee1812 Well....if your talking about the huananzhi board...I would skip those, seems kinda a hassle to get it working especially in overclocking. TF/F8/T8 should be the options. For the same price, I could get Ryzen 5600, B450 Asus Tuf (Bios flashback button) and 8gb x2 3200mhz CL16.
@@scprojekt1896 I got my x99 sli plus for free. Broken pcie slot tho. Doesn't have oc socket unfortunately so I can't get more than 4000mhz on the cache. I was able to get a rampage v extreme for $150. Cpus are cheap my 5960x was $60
I love that you only test games. The amount of my life I have wasted looking at bar graphs for Blender when ive never used the application is silly! Also a bit shocked to see my 5800X getting kneecapped by an i3 and an ageing i7!
But AMD compared the 1800x to the 6800k in Multi-Threaded Production workloads. NOT gaming. Everyone knew at the time that Intel's CPU's mopped the floor with Ryzen 1000 in gaming. Even a 6600k would give a 1800x a beat down. It was marketed towards people who did more than just game like streamers, and content creators ect. The whole point of it was to be a hybrid Main Stream Desktop and HEDT option for people who didn't want to spend boatloads of cash on INTEL's HEDT offerrings.
AMD compared 1800x to 6900K 8 core ($1000), not 6800K 6 core ($434). Even 1700 with OC at $330 was beating 6900K in MT, so it was a big win for creators, so they didn't needed to buy quad RAM or $1000 CPU for the same MT performance.
I’ve just got i7 5960x and asrock x99 taichi motherboard for £160 posted. (Update) It’s working very well got it overclocked to 4.5ghz on all core's and cache overclocked to 4.2ghz at 1.275v stable!! My ram using is Corsair vengeance 32gb 4x8 at 3200mhz, Also J batch chip so could go higher but haven’t tried.
this, I see 130w at full load on a 4.5ghz 6 core Xeon E5 1650 v2, the power savings with energy prices being what they are (especially in europe) makes crazy overclocks like that not optimal. Having to try to keep it cool is also a pain.
i dont why most people either ignore power consumption or just an addendum to the topic. that cpu overclocked and a good gaming gpu, the power draw is probably almost a kilowatt. with power at almost 25 cents per kilowatt, i wouldnt even consider even if was given the parts for free.
@@Alvyrre I agree. The performance sadly comes at a price especially if you live anywhere that isn't the USA, I've heard it's downright brutal for electricity in Europe atm.
Did you measure power efficiency? I believe 5960X with a stock TDP of 140W and when overclocked it would be around 200W at least. That's a lot when compared to 3700X when it's 65W stock and maybe 110W when highly overclocked or with PBO.
@@splashboi898 I do! I was running my 5960X rig at 4GHz but with power costs...it's now underclocked to 2.8GHz and minimum voltage. Dropped the wattage massively. Every little bit helps.
@@jabezhane that's .5ghz from the base frequency, not even the turbo. The i7 5960x is actually surprisingly power efficient. If you have a decent sample, you may only need 1.2 ish volts for 4.5ghz. I get similar performance to my old 11700k, which is why I returned that and kept this. It even uses less power too. You can undervolt the cpu what you did is underclocking which would drastically lower performance. There's no way at 2.5ghz it keeps up with a 6750 xt it's just impossible
Good video but i rather stick with b350 and ryzen 2600x, and then i am hoping to upgrade to the 5800x3d in few weeks and get a massive upgrade in performance comparable to intel 12900k in games, ryzen 1600 and 2600x can be found for the same price a just little more.
@@TheGoodOldGamer I need to play with what I have first and find those CPU limiting games/scene. Crysis 3 was always a good one inside the dome/train level.
@ 15:00 I do remember that and coming from the i7-6950X (4.4Ghz) in 32G quad 3200 Cl14 into a i9-10850K (5.2Ghz) same ram to 3600 in 2x16g (G.Skill Trident Z RGB), I felt the old 10 core kept up & some made up for it in quad vs dual Chan and it took the i9 pushing DDR-4000 to catch up. Bench #'s , IDK, but real world in Office or Adobe, X99 keeps up. However, it's easier to cool this little i9 130w on air compared to the big Broadwell-E 140w.
I don't find ths apples to apples comparison. Your compariing a HEDT part in the 5960x to a consumer mainstream part in the 3700x. You should do the comparison against a similar close equivalent in the Threadripper family. Also for the users who are more power consious in their machines, 5960x either at stock vs OC'd will consume way more power than the 3700x. Sure you get more performance with an OC, but it's not free you are paying for the performance by the increased electricty being consumed.
Hmmm... that would have to be tested. I have E5 1620v3 (which for all intents and purposes is an equivalent of 5820K but doesn't overclock nearly as much if you didn't win the silicon lottery - most cap out at around 4.2GHz) as of right now to stand in for my 5960X and at same OC (4.5) it's noticeably worse in single core performance in games.
All nice and professional. But I have other sources for this "usual" benches. What I would like to see is more multi player tests- they are totally different at times. Warzone, BFV, BF2042 etc...
I've been using the 5820K at 4.4 GHz with 4x16Gb ram at 2666 I had my eyes on the lookout for either a 5960x, Xeon 1660v3 or Xeon 1680v3. The prices of these CPUs were still pretty expensive until the recent weeks. After watching this and the previous video I went on eBay to see the current prices. I decided to buy the 5960x for around $65 USD. I know the Xeons may overclock better but I don't want the hassle of the ram not running at 2666+
Keep in mind that Xeons aren't overclocking nearly as well - while any 5960X or 5820K are pretty much guaranteed to hit 4.5 out of the box, rarely Xeons go higher than 4.2 - you can obviously be lucky with your purchase but there's no way to know until you test specific unit yourself.
What Vcore and cooling setup were you running? How were the temps in most games? You should do some productivity/synthetic benchmarks, those results could be interesting.
I have one since 2014, swapped 3 years ago to an xeon ES-1660 V3, overclocked at 4.5/4.0 DDR4 32 GB @3200Mhz @ CR 1T, cooled by an water loop. max temp in summer (38/40ºc) 66ºc. now with an rtx 3080 12gb, i noticed that i need a better CPU(mainly because of VR (rfactor2)), so, thinking now in the 5800x3D.
You need the right batch for the 5960X to reach 4.6/4.7Ghz and 3200Mhz memory speed. I had a chip made in Costa Rica (3429A714) which was running 4,7Ghz and 3200Mhz memory, which is more or less as high as these go unless you get a really golden one. In stresstest with a 280mm AIO it was running close to 90c.
Completely not surprised. I run an Ivy bridge Xeon e5-1680 V2 (8C/16T) OC's to 4.3ghz 24/7 (4.6ghz capable) equipped with 25MB of L3 cache (10mb more than i7 4960X plus 2 cores) running a RTX 2080Ti. It keeps up so well with modern CPU's at 4K max settings it is within 3-5FPS of the best CPUs out there, is competitive at 1440P and still very capable at 1080P though the lower the resolution with higher the refresh rates you're aiming for the more this chip handicaps you. Assuming you consider >90-100FPS a handicap. I even modded my bios to add M.2 boot support so I have most of the latest feature sets new systems do as well. Gaming wise it falls between around 1800x at 4.3ghz and 2700x at 4.6ghz. While I wouldn't buy one used now when I picked one up just before Zen 2 dropped for a hundred bucks or so 3 years ago as a drop in replacement for my i7 3930K because more games starting using all those cores/threads, it was a steal. It bought me enough time to wait for AM5 or raptor lake later this year. Probably longer but a decade of use out of the same motherboard has me itching for a newer build.
How hot are those Xeon? I am expecting to run at 200 watts at a minimum. If it is true, how those mobs can't stand such heat for so long? Paired whit a 2080 ti makes it 450 watts of pure heat.
@@goa141no6 My CPU never goes over 55C and my GPU never goes over 65C but I am water cooled on a custom parallel loops with 80mm thick rads 3x120mm that are completely outside of the case so the only waste heat the mobo has to deal with besides what it produces itself comes from the PSU, SSDs (4...2 SATA 2 NVME)/HDDs (5) and D5 water pumps. So my temps are pretty good across the board (7 case fans, 3x140mm, 2x120mm 2x 220mm) but I built my rig with temps in mind for a long term OC. I got ten years out of this mobo. I think that's enough to not complain. But y eah you're right. If you don't account for that heat you'll cook your motherboard's VRMs and caps real quick. I had an old i7 920's OC kill a mobo in a smaller ATX build that had poor ventilation in less then four years.
@@goa141no6 Actually I just checked at 4.3ghz it maxs at 184.7 watts on Cinebench 20...I took the OC to 4.6ghz though my guess is you'd be right. But with all that waste heat literally being dispelled on the outside of the case it really helps with internal case temps. I know an external rad doesn't look as clean as when it's mounted in your case but it is 1000% better on temps because you don't get radiating heat from the rads outer edge in the case nor from inefficient fan airflow back pumping hot air from your rad into your case PLUS your not limited on the size of your rad to fit your case. I seriously can't recommend an external radiator enough. Lower temps across the board and increased silence as fans can run slower while still doing their job.
Same here, was running a 1680v2 on an Asus IV Formula at 4.5ghz until I upgraded to a 3900x 5 years ago, also did the bios mod to boot from NVME. Just picked up a gigabyte x99 ud4p and a 1660v4 to play with.
I think you need to put a disclaimer or something because the pricing of the systems(mb/cpu/ram) on the release of Ryzen is vastly different to the pricing of the systems today, these both need to be shown not just today, easily find prices at the time if you look for them in youtube comparison videos /watch?v=Qmt8ND-FniI or the comments of them. Yes you're talking about the $ value you can get between them today but then you have to be clear that you can 't infer these price/performance to back when the One of the newer cpu's has released
Thank you for the video. Some days ago I got a X99 mobo for 60€ and a 5820k for 30€ to play around with pc hardware and today I just bought a 5960x for 50€... Then I found your video and you answered all questions in advanced 😂
I've got one in my basement with a Micro ATX Mobo with 2x8 GB 2133Mhz RAM :D Unbelievable that this CPU has performance that can come close to the new ones :D
I knew the chip would perform faster then a zen 2 CPU. I loved the x99 platform. I had 2 computers one had the 5960x and the other had 6950x. Used both for years before setting up a new system with 5900x. Wanted to try smart access memory with 6900xt. For gaming I don't notice much of a difference vs 1080ti with the 6950x @ 1440p. Thinking of building a new system with the 6950x and disabling 2 cores. Was waiting to see if you would try first since time for me is a commodity.
There is sometimes so much life left and also fun to be had with these older generations of hardware. But there is imo one important caveat esp if one has to oc in order to get this level of performance: IF one does not mind the increased power consumption and energy prices depending on where in the world one lives. In my case I would find such builds with this older gen hardware interesting, but no way near for a daily driver build and even less so if I were to build something for someone else. Like you mentioned a 12th gen i3, in 2022, is so much more of a logical and affordable choice to go for when building a 'new' system for anyone seeking a nice daily driver system.
Chris I like your videos I watch most of them and I listen to every TECHonomics podcast but your testing methodology is flawed in these videos in 2 ways. 1)You can't show Correlation or Causation with only 2 measurements(technically speaking as a Industrial Engineer)you can show possible trends but only 2 measurements doesn't show anything definitive. If you want an easy 3rd measurement just do Max Framerate hit while in play. The .1% lows make you more accurate but that isn't needed to show Correlations (THIS IS MORE OF A NITPICK BUT AS AN ENGINEER IT KINDA BUGS ME) 2)You can't make judgements using different settings. Why are the 5800X and 3700X not tested with the 2667 CL11. I assume you did this because it is common held knowledge that Ryzen loves Clock speed as long as it 1:1 on the Infinity Fabric but your tests bring up the idea of is this true. Would the low latency timings infact help the Ryzen CPU's or would it do nothing. Without measuring the CPU's using the same Ram and Same timings your testing is flawed(I mean technically you could argue if you are not testing on the same Motherboard Brand and Level it's flawed but I think thats going a bit far and would be more of the perfect being the enemy of the good enough) This is more important then my first point I also believe the 5960x is HEDT and ran quad Channel Memory and the 4 sticks of memory would be a benefit to it and Ryzen being Duel Channel and being known for having memory issues sometimes with 4 sticks because of it's memory controller could make it a problem. I've even seen others test and show that 2 sticks in Duel Channel on Ryzen is faster then 4 sticks on Ryzen even using the same memory. I hope you will take these things not as attacks or even criticisms but as points to make your testing better. Those numbers are surprising even with the flaws I pointed out so I’m not saying this video has no value or anything
I was interested too so built an X99 system and tested against my 9900k and the result is it almost matches stock 9900k when OCed to the max, within a few % but on memory I was limited to 3200 on the 5960x and I'm told it's good many can't do even 3200, 3600 didn't even post.
News flash: this 5960X OC-ed beats also anything from intel up until Alder Lake if it can match or beat Zen3, because Zen3 beats all of them until Alder Lake. So don't be a biased fanboi and don't omit important aspects and admit to everything, not only those in detriment of AMD. And if that's the case then intel has been stagnating for 7 years, again... But then again this is an extreme edition, so it's not meat to compete with the regular ones.
Hmm I think we need to do some retesting with a higher end GPU and at 1080p high settings, I think that 3060 is probably bottlenecking that overclocked 5800X even with DLSS enabled
Chris , I tried to get Steve at Gamers Nexus to do 5960X comparison for months but the new product affiliate sales sites will not cover salvage refurbish although Steve at Hardware UnBoxed appears to know why and has in GPU and is moving into CPU. Good job. You might want to report how you cooled 5960X. mb
@@PeterMarszalkowski the sweet spot for this cpu is 4.5, at 1.27v. More is stupid, because you will get much more heat, and watts, and not much more speed! 4.6 is maybe ok, more is not good.
You can run it at that speed in 1:1 on some CPU's, but most Zen 3 cap out at around 3800MHz. I asume that Chris tested both and was lucky enough to have a 5800x with a really good memory controller and didn't have any WHEA's or other issues at 4000MHz.
I like your enthusiasm about X99, but you are a bit late to the party as Eastern European TH-camrs have pointed this out for a while. I've been using Haswell-EP for some time now. As I tune my Arch-derivative with custom compiling a lot of packages, you'd be surprised what is still left in these older CPUs.
By the way, check the latest miyconst video, he did a GPU scaling benchmark comparison of a 10-core Haswell-EP vs. Alder Lake with several GPUs. tldr: Haswell-EP still has a lot to offer even with a 6800XT.
honestly, im really surprised by this. ive not checked out haswell-e in a fair few years because of board prices but just picked one up dirt cheap and would have thought it was way behind anything after first gen ryzen
Excellent find! I am as stunned at the results as you are. Great video. If I was looking for a purely gaming machine on a budget, I would be seriously considering this CPU.
The purchase price may be low on the cpu, but you'll lose the savings in electricity consumed by a cpu like the 5960x HEDT. Not to forget the extra heat being created if a person chooses to OC a 5960x
@@marcc5768 Good point. How much more could the 5960x consume relative to its competitors? If 100W more, that equates to paying ~$22 more a year for electricity if a person plays 4 hrs a day and electricity is 15 cents/hr.
@@dannydrew3685 craigslist. look for bundles, youll never find the mobo alone for a good deal. In a bundle I got the whole pc including a gtx 1080 for $200. Although the mobo was missing a pcie slot, ending up sellling that mobo for rampage extreme for 50 more.
Please, pretty please give those Zen3 systems another go with Resizable BAR enabled (forced if games aren't supported) to see if that adds another 10%+ lead in some games?
I have an odd question, that I hope you'll be able to shed some light on. I got my friend a 5960x and Gigabyte x99 UD3P motherboard. However, any time I try to overclock his CPU the overclock doesn't save. It just defaults back to stock setitings. Even a mild overclock like 4ghz with appropriate voltage and cooling. Have you heard of this problem/know any solutions to it? Thanks for your time!
I have been considering upgrading from an i7 5820k (4.3ghz) to a Ryzen 5 5600x or 7600x but I think I will wait. Interesting to read on user bench: 'The 5960X is only compatible with Intel's new X99 chipset which requires new DDR4 RAM and is currently far more expensive, with little consumer benefit, than DDR3' Much like DDR5 today.
@@gigapig23 You can try going for v3 Xeons for LGA2011v3 socket. They're pretty much an equivalent (except that they're not guaranteed to hit 4.5+ overclock) and there's plenty more since they're being pulled out en-masse from servers. For example equivalent of yours 5820K is E5-1620 v3 and it goes for like 20 quid... For 5960X equivalent you have to look for E5-1660 v3 which will be anywhere between 30-50... the caveat is that it may or may not be stable at 4.5GHz if you wanna take that gamble, it should hit 4.2 no problem though (I don't think I've ever seen one that didn't).
Non le 2700X est meilleur que le I7 5960X ... et consomme moins ! ( je ne parle pas de la version OC du I7 , qui consomme plus de 2x plus que le 2700X à coup sûr )
Just a generic comment, comparing plain FPS averages is not right (it is a common mistake made by tech reviewers). If you do math and calculate something you should understand what you are calculating, what does the result mean. To explain via a simplified example why this FPS average comparison is misleading: let's imagine that we compare 2 CPUs in 2 games. CPU A scores 50 FPS in game 1 and 190 FPS in game 2. CPU B scores 60 FPS in game 1 and 180 FPS in game 2. Their average FPS across the 2 games test is equal, 120 FPS. Thus they would be called equal performers in a similar video like this. However, if we take a look at the normalised results (meaning we set one CPU as baseline and calculate the relative performance of others compared to that baseline), let's say from the point of view of CPU A (meaning CPU A is 100%), then what we can see is that in game 1 CPU B is 20% faster than CPU A (CPU B=120%) while in game 2 CPU B is 5.3% slower than CPU A (CPU B=94.7%). I strongly doubt that anybody would accept the verdict that in overall the performances of CPU A and CPU B are equal. The inconvenient consequence of normalising is that the baseline must be always set to one member, and the relationship between the non-baseline members cannot be necessarily concluded from that data. If you are interested in any other particular member, you have to always re-calculate the normalisation, setting the baseline to that member. On the other hand, this plain average does not mean that thing how it is interpreted/understood in practice by ordinary people and can be quite distorted / misleading depending on the ranges/variances of the particular source data set and the magnitudes of data subsets. (This is why I ignore these 'aggregated average FPS of multiple games' charts in reviews. To show the original FPS numbers in the individual game charts is fine though, it gives credibility to the results.)
Yeah its a CPU that holds up decently with an overclock, but remember its 140w TDP at stock xD That overclock probably means CPU is sucking 200w plus at the wall
Wait so this thing is faster than my 9900k at 5.1ghz? Because last time I checked the 5800x stock is faster. Also when I checked an old digital foundry review with the 5960x it was losing to the 6700k in all the games they tested. Wtf
Yeah this cpu runs at a base of 3ghz and boosts to 3.5ghz on stock. And DDR4 launched on this platform so the speeds were only at 2133 at the time. Overclock the cores to 4.5 and the ram to 3200 quad channel (on par with DDR5 speeds), and this chip screams.
I found out about this CPU yesterday on the bottom of AMD Virtual super resolution page. At launch it was 1059$USD(MSRP). The Google demons then showed me your video mouahahah
Chris these results are indeed very interesting, unfortunately now (especially in europe, with what is happening: we could be without gas and electricity in the winter) power consuption is not irrelevant anymore, like at all, so... do you know how much does this overclocked 5960x consume?
@@TheGoodOldGamer Le I7 5960X même OC à 4,6 Ghz n'est absolument pas adapté à une RTX 3090. Beaucoup de watts gaspillé pour pas grand chose. Zen 2 fait tout aussi bien, avec beaucoup moins de consommation. Bref zen 2 est bien mieux au final.
@@daddyalpha2648 Really ?!. I have one, at the same speed, with 32 GB 3200/T1, raid 1 of 1 TB 4 HD. 1 M2, 2 SDD, 2 HDD each of 2TB, and an RTX 3080 12GB underclocked, but that still reach 324 Watts in some games. Power supply is an corsair 1000w, and 2 monitores, plus VR headset. AH, and more USB devices,maybe 5. not even 300w the CPU consume at that speed.
I guess I will have to test some of those games with my setup, looks like I choose right for a long lived platform. X99 FTW! i"9" 6950X, golden sample, OC upto 4.6Ghz. (14nm, 10C/20T, Broadwell-E) Rampage V Edition 10, LGA 2011-3. (OC socket with 2040 pins) Corsair DP SE Blackout, 32GB, DDR4 3200 CL14. (OC to 3600, Samsung B-Die) RTX 3080 Ti ROG Strix OC 12GB. I play at 4K, so the CPU would not be much of a problem anyways. Ha. 😃
@@MichaelJohnson-sb5si That's Ultimately the point I'm trying to get across. Even these OLD/Cheap CPUs can handle 120/144FPS. The brainwashing is STRONG out there right now.
Well this is his job, making videos for us to watch. I'm sure AdSense doesn't pay as much as you think, and nearly every tech channel has a community setup to join.
@@TheGoodOldGamer But even then most reviews just used stock CPU which benefit Ryzen more and Intel less. Zen1 was definitely slow for gaming. Also keep in mind after 5960x there is likely only 1 gen ipc with Skylake until 12th gen came out so if you had 5960x, you could ride the intel wave for 6 gen before you got a double gen increase.
$1060 SRP Testing at 720p or DLSS upscaled some perf mode is not legit plus using nVid does not allow SAM. Next a $1700 8950x vs less than $499 1800x, at 720p and DLSS upscale np SAM Compare vs a Threadrpper
![[UNCUT] The Loyal Pin ปิ่นภักดิ์ EP.4 (4/4)](http://i.ytimg.com/vi/d3l5rxl5CVo/mqdefault.jpg)
The HEDT Haswell chip had quad channel memory that gives the cpu access to a theoretical maximum memory bandwidth of about 85GB/s less a bit of over head. The Ringbus at 4.2ghz creates a pipe that connects the cpu cores to the installed memory and to the PCIe controller through a pipe that can potentially carry 134.4GB/s of data traffic. the 12100 Ringbus, if it runds at the same 4.2Ghz, creates a pipe of similar capacity to the 5950X but with the 3200 Ram can only get memory data at a maximum of 51.2GB/s less a bit of overhead. neither cpus can saturate the ring and there is a lot of extra capacity available for GPU data traffic that the CPU is generating while playing a game
On the Zen 1 and Zen 2 CPUs the infinity fabric performs the same function as the ring bus. The frequency is set to match the memory clock. if you have 3200 Ram installed, you have the Infinity Fabric running at 1600Mhz. The Ram can be delivered to the CPU at a maximum of 51.2GB/s just like the 12100. The infinity Fabric interconnects which are clocked the same as memory creates pipe that can carry 51.2GB/s. there is no extra overhead available if the ram is being read at max capacity to transfer data between CPU and GPU. Ram read capacity has to be compromised to allow the traffic to the GPU to share the same interconnect.
If you want to do something educational, try setting the Ringbus in the two intel boxes to a frequency that exactly matches the infinity fabric FCLK on the AMD machines. The memory frequency on the HEDT platform doesn't matter cause u have quad channel memory installed an an abundance of memory that is limited to 51.2GB/s anyway. All of the dual channel systems should be able to run 3200 DDR4 as the lowest common denominator so select that and then set the ringbus to 1.6GHz so it matches the IF pipes. I think that you may find the changes in performance on the intel systems, compared to the Ryzen systems interesting. I would also suggest running cinebench on each system. It mostly runs out of the cache so it is a good measure for creating a baseline of the relative ipc of each CPU without being too much by the memory speed. You should see the performance variations in games look similar to the variations in the cinebench scores rather than what we are all used to seeing.
The test will show how hobbled Ryzen was with zen1 and how it improved over the generations and also demonstrate why Ryzen traditionally struggled with gaming workloads before and not now with the 5800x. The 5800x has double the infinity fabric interconnect capacity so it behaves more like an intel cpu but It can also be used to show that the main improvement in performance in Ryzen comes primarily from the bigger pipe created by the faster FCLK, Faster Ram certainly provides a bit of the boost to performance but it is not, as everyone seems to assume, from latency and brute strength of the faster silicon in the 3800/4000 ddr4 chips.
Clearly this is a test for science, not a recommendation to daily your rig with ringbus set well below the default settings.
This is the best comment I’ve ever seen on a tech tubers video. Kudos to you
Fascinating! I clearly don't understand much of the mechanics of this but I get the concepts. I think! Thanks.
@@informedchoice2249 In conceptual terms, a Ring bus is like a 4 lane ringroad highway that connects the cores to the memory and the gpu/pcie bus with on/off ramps at each location. The data can go in either direction to get to the other destination without congestion.
Infinity fabric is built like a two lane road system with an intersection in the middle. the io die acts like the traffic lights managing the traffic. there is only the one path between each destination going through the central crossroads. Like in the roads, the traffic needs to be managed to prevent collisions at the intersecton which leads to some traffic congestion at periods of heavy traffic
The Extreme series CPUs are always cheap, it's the motherboards that cost a ton, especially in smaller countries.
Yeah , but I'd say , if you know how to utilize the features , you'd not regret buying the motherboard ..
@@yasirrakhurrafat1142 I know hot to utilize even server grade motherboards to their fullest potential, that doesn't mean that if I was building a gaming PC be first or second hand I'd care for them. We are talking pure gaming here.
I'd rather get a ryzen 5 3600 with a B450/550 board second hand for a total of 200-250$ than a I7 5960x for 80$(here it's 110-130$) and a X99 motherboard for close to 250-350$.
Also the E5 2620v3 and 2630 V3 go for like 20-40$, but they can't be overclocked as easily and as much.
@@badass6300
Irrelevant but , isn't the x99 socket setup considered HEDT ?
And yeah you're right , low cpu intensive titles run well on max on even 12100 . While providing high performance .
I was just trying to engage you to the endless possibilities of the x99/c602 chipset , for the budget .
Just imagine getting the 2699v3 oc it to the max , and using it for multiple gaming sessions locally or remote , through VMs or Multiseat .
Rather than building 8 boxes , you'd build have to build 2 , for medium spec Gaming on each instance .
@@yasirrakhurrafat1142 hi, did u mention vms ? what setup is good for multiple gaming sessions on the same pc ? multiboxing, do u know
@@laughingcross8108 its kind of a individual preference , as to how much performance you want per vm .
But if you're gaming , two users/vms , top demanding games , you should be good with an i7 12th gen or a ryzen 5800 , with 32gb of ram and 3050s( per vm ) or 3060s( per vm ) .
For absolute above 60 fps performance in 1080p high in all games , 1440p high 60 fps in some .
I wasn’t surprised to see possibly that the Intel 5960x would at least come close to Zen 2 ….. but Zen 3 was a mind blown 🤯🤯🤯 situation for me! Absolutely lovin 💕 this series of tweaking memory timings on different generations of AMD / Intel CPUs’. Truely eye opening a number of these results 🫠🤩🤩
Back in the day I had a rig with the 5960X with a pair of GTX 980Ti's and it was a beast for it's time. The 5960X is still a beast of a CPU even against some of the newer stuff.
If this isn't a full fledged endorsement of Overclocking then I don't know what is! These results are mind-blowing and gives me hope that people in the future will learn how important properly tuning your system can be. Great video!
that performance doesn't come for free! Any numbers for power consumption?
I recently upgraded my i7 5820k system to a 5960x, overclocked. I think I'll be able to get another few years out of this machine. That will make it the first computer I get a full decade out of without upgrading the motherboard. Intel really outdid themselves on this CPU for the time.
well, it was 450 Euros and 1000 Euros for the board and CPU. DDR4 was more expensive than even DDR5 was. It was Xeon processors with I7 names.
Same here my friend. Rocking the ASUS Rampage V E10 mobo with 1000w H1000x psu and 3200 Corsair RGB RAM with a FE 3090 . I run a 5k screen and maxing out 150-200% MSAA or Full ray tracing without DLSS garbage, I rock 60fps all day and that is all I will ever ask for.
I've run a 5960x for 8 years but finally replaced it.
@@lewisner Noticeably different ?
@Vinson Smith I mainly use it for video games and it isn't that much different. The 5960x was way over spec for what I needed but I was keen then and I liked to have the best.
There is a reason these CPU's were over $1,000 6 years ago. That 10-core was my dream CPU for a long time. I'm on a 8700k which is fine tho.
le 8700K est bien meilleur que ce I7 5960X ... il ne faut pas oublier qu'il faut OC à l'extrême ce cpu, et que ça engendre une consommation de plus de 300W
It's 8 cores/16 threads. I have a 9900K and it only has 16 MB of L3 cache, while the 5960X has 20 MB. The size of the L3 cache really helps with gaming performance.
@@tenseur2 no its not much better not even close lol. If the 5960x beats zen3 and alderlake how can the 8700k be faster than? 🙄 Where's your logic? 😂
so glad i picked up the i3 12100f ! its such a good little chip just for gaming!
Im not amazed for a simple reason that I was using i7-3930K @4500Mhz + DDR3 4x4 2400 11-13-12-20-1 for about 9 years, until I bought i9-12900K because.... reasons... I guess, but the 3930K didn't gave me any doubts that I need a new platform, and it was so stable and strong still. You need to check it next. ! :) It is a legendary platform (X79). Good work dude ! ;)
There was a time a few years ago when one could get top end X79 mbds very cheap, I bought a bunch of them, mainly the ASUS P9X79/E WS. Best deal I bought was an ASUS Rampage IV Extreme, 3930K C2 and 16GB/2400 RAM for just 200 UKP (it formed the basis of a gaming PC I built for a friend, though the CPU was swapped out for a 4930K).
I moved from my x99 5960x to x299 and 7900x, more cores, a bit higher frequencies, but shame there's lesser l3 cache though on skylake-x chip.. Looking to get 9900x or even 10900x in time the price drops even more. Going to delid my 7900x in the mean time.. It's awesome these chips are now so affordable, I'm not even looking at current platforms, my HEDT is still going strong and will be good for another 5+ years easy.
There problem back then was that the Core i7 5960X was a $999 price and the motherboards was expensive too. You have to overclock it to really compete, which will reduce the live span of an already old CPU released on 2014.
On today prices is a great project to run until it die, but was not possible at the R7 5800 release date, since the Core i7 5960X used was at ~$500-600 price on a expensive platform.
The i3 12100 still in my opinion the best pure gaming machine for mid-high gaming target. But what I want too see is benchmarks where the machine have like a chrome with 20+ taps, two monitors, playing twich or youtube and other background application running at the same time you game, there is where Ryzen shine. My use case will never be just the game running.
The Xeon alternative has always been the better buy for years now. 1660 v3
@@lmaoitsmee1812 if your talking about live span when overclock....it always the motherboard that will be stressed more than the cpu. Between the two, I would say the motherboard wouldn't even hold up unless it's a really high end one. The cpu will always survive. Rarely dead, really rare. If you suspect the cpu is dead, it always the motherboard that is the problem. X99 is kinda still expensive, I would rather go for 5600X regardless, newer and easy to get 😂😂
12100 is the best bang for the buck, period!
@@scprojekt1896 x99 is expensive for those that do not know where to look. god bless.
@@lmaoitsmee1812 Well....if your talking about the huananzhi board...I would skip those, seems kinda a hassle to get it working especially in overclocking. TF/F8/T8 should be the options. For the same price, I could get Ryzen 5600, B450 Asus Tuf (Bios flashback button) and 8gb x2 3200mhz CL16.
@@scprojekt1896 I got my x99 sli plus for free. Broken pcie slot tho. Doesn't have oc socket unfortunately so I can't get more than 4000mhz on the cache. I was able to get a rampage v extreme for $150. Cpus are cheap my 5960x was $60
I love that you only test games. The amount of my life I have wasted looking at bar graphs for Blender when ive never used the application is silly!
Also a bit shocked to see my 5800X getting kneecapped by an i3 and an ageing i7!
But AMD compared the 1800x to the 6800k in Multi-Threaded Production workloads. NOT gaming. Everyone knew at the time that Intel's CPU's mopped the floor with Ryzen 1000 in gaming. Even a 6600k would give a 1800x a beat down. It was marketed towards people who did more than just game like streamers, and content creators ect. The whole point of it was to be a hybrid Main Stream Desktop and HEDT option for people who didn't want to spend boatloads of cash on INTEL's HEDT offerrings.
AMD compared 1800x to 6900K 8 core ($1000), not 6800K 6 core ($434). Even 1700 with OC at $330 was beating 6900K in MT, so it was a big win for creators, so they didn't needed to buy quad RAM or $1000 CPU for the same MT performance.
@@steamstories1279 Yeah, my bad on that, but atleast you understand the point I was trying to make.
I’ve just got i7 5960x and asrock x99 taichi motherboard for £160 posted. (Update) It’s working very well got it overclocked to 4.5ghz on all core's and cache overclocked to 4.2ghz at 1.275v stable!! My ram using is Corsair vengeance 32gb 4x8 at 3200mhz, Also J batch chip so could go higher but haven’t tried.
With Engery prices the way they are, I'll stick with a 5800x, the power draw of that Intel CPU is crazy at those clocks!
this, I see 130w at full load on a 4.5ghz 6 core Xeon E5 1650 v2, the power savings with energy prices being what they are (especially in europe) makes crazy overclocks like that not optimal. Having to try to keep it cool is also a pain.
i dont why most people either ignore power consumption or just an addendum to the topic. that cpu overclocked and a good gaming gpu, the power draw is probably almost a kilowatt. with power at almost 25 cents per kilowatt, i wouldnt even consider even if was given the parts for free.
@@Alvyrre I agree. The performance sadly comes at a price especially if you live anywhere that isn't the USA, I've heard it's downright brutal for electricity in Europe atm.
Are people poor who cares about power draw lol
@@Mvp-AngelOfGod take a look at european energy prices and then compare that with their taxation rate, it's a big deal.
Hey Chris, got fond memories of Haswell / Devils Canyon, that architecture was great for its time, and lol..today.
Great Video.
Did you measure power efficiency? I believe 5960X with a stock TDP of 140W and when overclocked it would be around 200W at least. That's a lot when compared to 3700X when it's 65W stock and maybe 110W when highly overclocked or with PBO.
Who cares
@@splashboi898 I do! I was running my 5960X rig at 4GHz but with power costs...it's now underclocked to 2.8GHz and minimum voltage. Dropped the wattage massively. Every little bit helps.
@@jabezhane 2.8ghz 😱
@@splashboi898 It was 2.5 but it still does all the work and gaming I do. Still got quad ddr4 and a 6700XT....
@@jabezhane that's .5ghz from the base frequency, not even the turbo. The i7 5960x is actually surprisingly power efficient. If you have a decent sample, you may only need 1.2 ish volts for 4.5ghz. I get similar performance to my old 11700k, which is why I returned that and kept this. It even uses less power too. You can undervolt the cpu what you did is underclocking which would drastically lower performance. There's no way at 2.5ghz it keeps up with a 6750 xt it's just impossible
Good video but i rather stick with b350 and ryzen 2600x, and then i am hoping to upgrade to the 5800x3d in few weeks and get a massive upgrade in performance comparable to intel 12900k in games, ryzen 1600 and 2600x can be found for the same price a just little more.
Really need to test this myself. Seems pretty crazy increase in performance. I have 5820K which is 6C but that should not effect gaming performance.
Yeah man it’s a lot of fun. The difference in cache will make a difference though
@@TheGoodOldGamer I need to play with what I have first and find those CPU limiting games/scene. Crysis 3 was always a good one inside the dome/train level.
Also on a good note. you can't be updated to Win11 with I7-5960x.
@ 15:00 I do remember that and coming from the i7-6950X (4.4Ghz) in 32G quad 3200 Cl14 into a i9-10850K (5.2Ghz) same ram to 3600 in 2x16g (G.Skill Trident Z RGB), I felt the old 10 core kept up & some made up for it in quad vs dual Chan and it took the i9 pushing DDR-4000 to catch up. Bench #'s , IDK, but real world in Office or Adobe, X99 keeps up. However, it's easier to cool this little i9 130w on air compared to the big Broadwell-E 140w.
If you OC TDP is merely a suggestion. My 5960X at 4.5 was hitting 200W territory...
I don't find ths apples to apples comparison. Your compariing a HEDT part in the 5960x to a consumer mainstream part in the 3700x. You should do the comparison against a similar close equivalent in the Threadripper family. Also for the users who are more power consious in their machines, 5960x either at stock vs OC'd will consume way more power than the 3700x. Sure you get more performance with an OC, but it's not free you are paying for the performance by the increased electricty being consumed.
Thats why 5820K can be nominated as best released CPU overall
Highest longevity and great price/perf
Hmmm... that would have to be tested. I have E5 1620v3 (which for all intents and purposes is an equivalent of 5820K but doesn't overclock nearly as much if you didn't win the silicon lottery - most cap out at around 4.2GHz) as of right now to stand in for my 5960X and at same OC (4.5) it's noticeably worse in single core performance in games.
All nice and professional. But I have other sources for this "usual" benches.
What I would like to see is more multi player tests- they are totally different at times. Warzone, BFV, BF2042 etc...
I've been using the 5820K at 4.4 GHz with 4x16Gb ram at 2666
I had my eyes on the lookout for either a 5960x, Xeon 1660v3 or Xeon 1680v3.
The prices of these CPUs were still pretty expensive until the recent weeks.
After watching this and the previous video I went on eBay to see the current prices.
I decided to buy the 5960x for around $65 USD.
I know the Xeons may overclock better but I don't want the hassle of the ram not running at 2666+
I have an xeon 1660v3 at 4.5/4.0 @3200/T1 32GB it can run at 4.6 all day, but only in the winter. (Water cooled)
Keep in mind that Xeons aren't overclocking nearly as well - while any 5960X or 5820K are pretty much guaranteed to hit 4.5 out of the box, rarely Xeons go higher than 4.2 - you can obviously be lucky with your purchase but there's no way to know until you test specific unit yourself.
What Vcore and cooling setup were you running? How were the temps in most games?
You should do some productivity/synthetic benchmarks, those results could be interesting.
I have one since 2014, swapped 3 years ago to an xeon ES-1660 V3, overclocked at 4.5/4.0 DDR4 32 GB @3200Mhz @ CR 1T, cooled by an water loop. max temp in summer (38/40ºc) 66ºc. now with an rtx 3080 12gb, i noticed that i need a better CPU(mainly because of VR (rfactor2)), so, thinking now in the 5800x3D.
You need the right batch for the 5960X to reach 4.6/4.7Ghz and 3200Mhz memory speed. I had a chip made in Costa Rica (3429A714) which was running 4,7Ghz and 3200Mhz memory, which is more or less as high as these go unless you get a really golden one.
In stresstest with a 280mm AIO it was running close to 90c.
The real star here is the i3 12100 at $100 its really good and gets you started on a modern platform.
It almost pains me to see how many x99 motherboards are on eBay without CPU socket covers.
Should have included non OC and tuned result to see how much people are missing out by not OC these chips.
I love content like this. I'm pretty reliant on current level of I/o, so something like x99 just doesn't cut it.. but it's still fun to think about
Completely not surprised. I run an Ivy bridge Xeon e5-1680 V2 (8C/16T) OC's to 4.3ghz 24/7 (4.6ghz capable) equipped with 25MB of L3 cache (10mb more than i7 4960X plus 2 cores) running a RTX 2080Ti. It keeps up so well with modern CPU's at 4K max settings it is within 3-5FPS of the best CPUs out there, is competitive at 1440P and still very capable at 1080P though the lower the resolution with higher the refresh rates you're aiming for the more this chip handicaps you. Assuming you consider >90-100FPS a handicap. I even modded my bios to add M.2 boot support so I have most of the latest feature sets new systems do as well. Gaming wise it falls between around 1800x at 4.3ghz and 2700x at 4.6ghz. While I wouldn't buy one used now when I picked one up just before Zen 2 dropped for a hundred bucks or so 3 years ago as a drop in replacement for my i7 3930K because more games starting using all those cores/threads, it was a steal. It bought me enough time to wait for AM5 or raptor lake later this year. Probably longer but a decade of use out of the same motherboard has me itching for a newer build.
How hot are those Xeon? I am expecting to run at 200 watts at a minimum. If it is true, how those mobs can't stand such heat for so long? Paired whit a 2080 ti makes it 450 watts of pure heat.
@@goa141no6 My CPU never goes over 55C and my GPU never goes over 65C but I am water cooled on a custom parallel loops with 80mm thick rads 3x120mm that are completely outside of the case so the only waste heat the mobo has to deal with besides what it produces itself comes from the PSU, SSDs (4...2 SATA 2 NVME)/HDDs (5) and D5 water pumps. So my temps are pretty good across the board (7 case fans, 3x140mm, 2x120mm 2x 220mm) but I built my rig with temps in mind for a long term OC. I got ten years out of this mobo. I think that's enough to not complain. But y eah you're right. If you don't account for that heat you'll cook your motherboard's VRMs and caps real quick. I had an old i7 920's OC kill a mobo in a smaller ATX build that had poor ventilation in less then four years.
@@wheelchairskater Now i see the usefulness of water cooler over air cooler, looks like 400 series could be only water cooler from the 4080 and up.
@@goa141no6 Actually I just checked at 4.3ghz it maxs at 184.7 watts on Cinebench 20...I took the OC to 4.6ghz though my guess is you'd be right. But with all that waste heat literally being dispelled on the outside of the case it really helps with internal case temps. I know an external rad doesn't look as clean as when it's mounted in your case but it is 1000% better on temps because you don't get radiating heat from the rads outer edge in the case nor from inefficient fan airflow back pumping hot air from your rad into your case PLUS your not limited on the size of your rad to fit your case. I seriously can't recommend an external radiator enough. Lower temps across the board and increased silence as fans can run slower while still doing their job.
Same here, was running a 1680v2 on an Asus IV Formula at 4.5ghz until I upgraded to a 3900x 5 years ago, also did the bios mod to boot from NVME. Just picked up a gigabyte x99 ud4p and a 1660v4 to play with.
I think you need to put a disclaimer or something because the pricing of the systems(mb/cpu/ram) on the release of Ryzen is vastly different to the pricing of the systems today, these both need to be shown not just today, easily find prices at the time if you look for them in youtube comparison videos /watch?v=Qmt8ND-FniI or the comments of them. Yes you're talking about the $ value you can get between them today but then you have to be clear that you can 't infer these price/performance to back when the One of the newer cpu's has released
All the security issues these old Intel chip have are a big problem. These results are insane, but I rather go with a newer platform.
Damn, had forgotten CPUs got sandbagged by security updates. Probably goes a long way in explaining the discrepancies here.
@@dickjohnson8036 this cpu also had security update via OS and via bios. Is not that.
I don't even think ppl tested gaming vs The 1800X because they were so fixed on Multi Threaded Performance with a $500 chip vs a $1000 chip.
what was the wattage pulled by the i7? during these tests and temps
Great video! It would be really interesting to see how i7-5820k compares as it can be had for even cheaper
Thank you for the video. Some days ago I got a X99 mobo for 60€ and a 5820k for 30€ to play around with pc hardware and today I just bought a 5960x for 50€... Then I found your video and you answered all questions in advanced 😂
I've got one in my basement with a Micro ATX Mobo with 2x8 GB 2133Mhz RAM :D
Unbelievable that this CPU has performance that can come close to the new ones :D
I'm still rocking my 5960x (and ASUS 1080 TI) in 2023.
I knew the chip would perform faster then a zen 2 CPU. I loved the x99 platform. I had 2 computers one had the 5960x and the other had 6950x. Used both for years before setting up a new system with 5900x. Wanted to try smart access memory with 6900xt. For gaming I don't notice much of a difference vs 1080ti with the 6950x @ 1440p. Thinking of building a new system with the 6950x and disabling 2 cores. Was waiting to see if you would try first since time for me is a commodity.
There is sometimes so much life left and also fun to be had with these older generations of hardware.
But there is imo one important caveat esp if one has to oc in order to get this level of performance: IF one does not mind the increased power consumption and energy prices depending on where in the world one lives. In my case I would find such builds with this older gen hardware interesting, but no way near for a daily driver build and even less so if I were to build something for someone else.
Like you mentioned a 12th gen i3, in 2022, is so much more of a logical and affordable choice to go for when building a 'new' system for anyone seeking a nice daily driver system.
Even Bloomfield still got game. And that's a 2008 CPU.
Chris I like your videos I watch most of them and I listen to every TECHonomics podcast but your testing methodology is flawed in these videos in 2 ways.
1)You can't show Correlation or Causation with only 2 measurements(technically speaking as a Industrial Engineer)you can show possible trends but only 2 measurements doesn't show anything definitive. If you want an easy 3rd measurement just do Max Framerate hit while in play. The .1% lows make you more accurate but that isn't needed to show Correlations (THIS IS MORE OF A NITPICK BUT AS AN ENGINEER IT KINDA BUGS ME)
2)You can't make judgements using different settings. Why are the 5800X and 3700X not tested with the 2667 CL11. I assume you did this because it is common held knowledge that Ryzen loves Clock speed as long as it 1:1 on the Infinity Fabric but your tests bring up the idea of is this true. Would the low latency timings infact help the Ryzen CPU's or would it do nothing. Without measuring the CPU's using the same Ram and Same timings your testing is flawed(I mean technically you could argue if you are not testing on the same Motherboard Brand and Level it's flawed but I think thats going a bit far and would be more of the perfect being the enemy of the good enough) This is more important then my first point
I also believe the 5960x is HEDT and ran quad Channel Memory and the 4 sticks of memory would be a benefit to it and Ryzen being Duel Channel and being known for having memory issues sometimes with 4 sticks because of it's memory controller could make it a problem. I've even seen others test and show that 2 sticks in Duel Channel on Ryzen is faster then 4 sticks on Ryzen even using the same memory.
I hope you will take these things not as attacks or even criticisms but as points to make your testing better. Those numbers are surprising even with the flaws I pointed out so I’m not saying this video has no value or anything
Keep up the good work. X58 Westmere vs Zen1 is a pretty good battle too 6/12 vs 6/16
this cpu sounds like a good chip for a home lab server for proxmox.
I'm still rocking the 5960X. It's not as tuned with CL11 memory, but even paired with a 4090 - I'm still gaming happily at stupid settings and perf 😅
Surprise surprise Intel's silicon is ultimately top notch and remains competitive even with older nodes.
It would be interesting to see a similar comparison but against an overclocked 9900k since that one can draw as much power as the 5960x
I was interested too so built an X99 system and tested against my 9900k and the result is it almost matches stock 9900k when OCed to the max, within a few % but on memory I was limited to 3200 on the 5960x and I'm told it's good many can't do even 3200, 3600 didn't even post.
News flash: this 5960X OC-ed beats also anything from intel up until Alder Lake if it can match or beat Zen3, because Zen3 beats all of them until Alder Lake. So don't be a biased fanboi and don't omit important aspects and admit to everything, not only those in detriment of AMD.
And if that's the case then intel has been stagnating for 7 years, again... But then again this is an extreme edition, so it's not meat to compete with the regular ones.
Hmm I think we need to do some retesting with a higher end GPU and at 1080p high settings, I think that 3060 is probably bottlenecking that overclocked 5800X even with DLSS enabled
It's bottlenecking all of the CPUs, so the point is moot!
Good salvage coverage pushing that DDR 2667? Yea, I think so. 5960X ring bus, whole circuit fabricated at 22 nm is a very clean lay up. mb
Chris , I tried to get Steve at Gamers Nexus to do 5960X comparison for months but the new product affiliate sales sites will not cover salvage refurbish although Steve at Hardware UnBoxed appears to know why and has in GPU and is moving into CPU. Good job. You might want to report how you cooled 5960X. mb
@@mikebruzzone9570 4.7ghz 95c ^^ too many core treads wrong cpu ^^ sorry doesn't last long ^^
@@PeterMarszalkowski Thanks for 5960X pushed data. mb
@@PeterMarszalkowski the sweet spot for this cpu is 4.5, at 1.27v. More is stupid, because you will get much more heat, and watts, and not much more speed! 4.6 is maybe ok, more is not good.
How'd you get the RAM on the 5800x to run at 4000MHz, I thought RAM on RYZEN couldn't go past 3800MHz?
You can run it at that speed in 1:1 on some CPU's, but most Zen 3 cap out at around 3800MHz. I asume that Chris tested both and was lucky enough to have a 5800x with a really good memory controller and didn't have any WHEA's or other issues at 4000MHz.
What about the power consumption and the efficiency
tdp is in reviews, and efficiency is 0% all watts are wasted as heat. only psu has efficiency.
I like your enthusiasm about X99, but you are a bit late to the party as Eastern European TH-camrs have pointed this out for a while. I've been using Haswell-EP for some time now. As I tune my Arch-derivative with custom compiling a lot of packages, you'd be surprised what is still left in these older CPUs.
By the way, check the latest miyconst video, he did a GPU scaling benchmark comparison of a 10-core Haswell-EP vs. Alder Lake with several GPUs. tldr: Haswell-EP still has a lot to offer even with a 6800XT.
What about power consumption?
Thank you for this & everythin' you do on this channel. appreciate it.
Great job!! You are doing AWESOME!!
honestly, im really surprised by this. ive not checked out haswell-e in a fair few years because of board prices but just picked one up dirt cheap and would have thought it was way behind anything after first gen ryzen
Not bad for a 22nm part
Only more evidence intel was sitting doing nothing for a decade.
Excellent find! I am as stunned at the results as you are. Great video. If I was looking for a purely gaming machine on a budget, I would be seriously considering this CPU.
The purchase price may be low on the cpu, but you'll lose the savings in electricity consumed by a cpu like the 5960x HEDT. Not to forget the extra heat being created if a person chooses to OC a 5960x
@@marcc5768 Good point. How much more could the 5960x consume relative to its competitors? If 100W more, that equates to paying ~$22 more a year for electricity if a person plays 4 hrs a day and electricity is 15 cents/hr.
@@bgtubber At that level of OC is more than 100W
@@lupintheiii3055 Interesting. I didn't know that. Thanks!
Goes to show how memory bottlenecked modern CPUs are.
awesome content, please keep it coming!!
Do the test with a RX 6900, RX 6950, 3090, or 3090 ti and I bet the results are different
You're welcome to send all of those for testing. BTW I did the 3090. Same Results.
@@TheGoodOldGamer lol, is eBay the best place to get the mobo’s for that chip? It’s over $200 everywhere else.
@@dannydrew3685 they where super expensive when new, 500 Euros for the good ones.
@@dannydrew3685 craigslist. look for bundles, youll never find the mobo alone for a good deal. In a bundle I got the whole pc including a gtx 1080 for $200. Although the mobo was missing a pcie slot, ending up sellling that mobo for rampage extreme for 50 more.
yes, quad channel ddr4 bandwith and chonkee L3 cache does it.
Please, pretty please give those Zen3 systems another go with Resizable BAR enabled (forced if games aren't supported) to see if that adds another 10%+ lead in some games?
I have an odd question, that I hope you'll be able to shed some light on. I got my friend a 5960x and Gigabyte x99 UD3P motherboard. However, any time I try to overclock his CPU the overclock doesn't save. It just defaults back to stock setitings. Even a mild overclock like 4ghz with appropriate voltage and cooling. Have you heard of this problem/know any solutions to it? Thanks for your time!
Is the RAM on the i7-5960X System configured in Quad-Channel or are all CPUs including the i7 utilizing Dual-Channel RAM?
I have been considering upgrading from an i7 5820k (4.3ghz) to a Ryzen 5 5600x or 7600x but I think I will wait.
Interesting to read on user bench:
'The 5960X is only compatible with Intel's new X99 chipset which requires new DDR4 RAM and is currently far more expensive, with little consumer benefit, than DDR3'
Much like DDR5 today.
Pick yourself up a 5960X for not much and have a play around with it till you actually get around to upgrading.
@@jabezhane Thank you. I've looked on Ebay UK and they are rare or around £100 but I will keep looking.
@@gigapig23 I sold my 5820K after I got one a couple of years ago to offset the cost.
@@gigapig23 You can try going for v3 Xeons for LGA2011v3 socket. They're pretty much an equivalent (except that they're not guaranteed to hit 4.5+ overclock) and there's plenty more since they're being pulled out en-masse from servers. For example equivalent of yours 5820K is E5-1620 v3 and it goes for like 20 quid...
For 5960X equivalent you have to look for E5-1660 v3 which will be anywhere between 30-50... the caveat is that it may or may not be stable at 4.5GHz if you wanna take that gamble, it should hit 4.2 no problem though (I don't think I've ever seen one that didn't).
should I buy an i7 5960x then? am upgrading from R7 2700x :)
no it be cheaper to get a 5800x or 5800x3D because the chips might be cheap but the Mobo's are not.
12100 for gaming man. Nothing better for the money.
Non le 2700X est meilleur que le I7 5960X ... et consomme moins ! ( je ne parle pas de la version OC du I7 , qui consomme plus de 2x plus que le 2700X à coup sûr )
@@TheGoodOldGamer Oui et non. Sur le court terme oui, sur du moyen terme non ... car limité à 8 threads.
Nope. 5800X.
Just a generic comment, comparing plain FPS averages is not right (it is a common mistake made by tech reviewers). If you do math and calculate something you should understand what you are calculating, what does the result mean.
To explain via a simplified example why this FPS average comparison is misleading: let's imagine that we compare 2 CPUs in 2 games. CPU A scores 50 FPS in game 1 and 190 FPS in game 2. CPU B scores 60 FPS in game 1 and 180 FPS in game 2. Their average FPS across the 2 games test is equal, 120 FPS. Thus they would be called equal performers in a similar video like this. However, if we take a look at the normalised results (meaning we set one CPU as baseline and calculate the relative performance of others compared to that baseline), let's say from the point of view of CPU A (meaning CPU A is 100%), then what we can see is that in game 1 CPU B is 20% faster than CPU A (CPU B=120%) while in game 2 CPU B is 5.3% slower than CPU A (CPU B=94.7%). I strongly doubt that anybody would accept the verdict that in overall the performances of CPU A and CPU B are equal.
The inconvenient consequence of normalising is that the baseline must be always set to one member, and the relationship between the non-baseline members cannot be necessarily concluded from that data. If you are interested in any other particular member, you have to always re-calculate the normalisation, setting the baseline to that member.
On the other hand, this plain average does not mean that thing how it is interpreted/understood in practice by ordinary people and can be quite distorted / misleading depending on the ranges/variances of the particular source data set and the magnitudes of data subsets. (This is why I ignore these 'aggregated average FPS of multiple games' charts in reviews. To show the original FPS numbers in the individual game charts is fine though, it gives credibility to the results.)
ah, spreading the good will of LGA2011-3
I was happy w/ the i7-5960X. But, how much fun OC'n can the 6850K or i7-5930K w/ 2 less core overhead be?
What about the power consumption comparing the old i7 to the i3 / Ryzen 7?
Wow I didn't expect this, too bad Intel stopped making HEDT.
Intel HEDT is back in 2023 with Sapphire Rapids/W790, but... the motherboards are upwards of £900, and they require registered DDR5.
How on Earth did you manage to bump Ringbus to 4.2? O_o
Have you tested 3600-3800 cl14 instead of 4000 c16 on the Ryzen 5800x?
what is the model number of that Sony TV behind him again? Isn't that fabled to be the best CRT ever made for gaming?
Why did X99 cpus use ring bus, but X299 cpus like 10980XE use mesh?😊
This is so awesome dude.. the bdie tuning and quad channel memory paired with ringbus is hard to beat. And for $80.. lol
What's with the quality of the video?
You wont be disappointed pairing this with a 3090
stock results would be nice, for people who buy used workstations they can't overclock their xeons on
Yeah its a CPU that holds up decently with an overclock, but remember its 140w TDP at stock xD
That overclock probably means CPU is sucking 200w plus at the wall
At 4.5 ghz close to 500 watts ....
@@daddyalpha2648 More than 300W easy !!
@@tenseur2 like I wrote at 4.5 ghz close to 500 watts. 400 and something. So it already is more than your 300 watts 🙄
@@daddyalpha2648 L'efficacité énergétique de ce I7 est catastrophique
@@tenseur2 why did you reply in french now but previously in english? I dont understand french 😂
Are you sure that 5800x is 100% stable at 2000 fclk? No Whea errors?
Wait so this thing is faster than my 9900k at 5.1ghz? Because last time I checked the 5800x stock is faster. Also when I checked an old digital foundry review with the 5960x it was losing to the 6700k in all the games they tested. Wtf
it was not overclocked, or not using DDR4 at 3200, is a world of difference. And also the ring bus overclock.
Yeah this cpu runs at a base of 3ghz and boosts to 3.5ghz on stock. And DDR4 launched on this platform so the speeds were only at 2133 at the time.
Overclock the cores to 4.5 and the ram to 3200 quad channel (on par with DDR5 speeds), and this chip screams.
i have a 4790, should i get a 5960x?? this is crazy!
This video makes the Yes! Man proud.
can you do a 6900k comp?
id love to see how mines still doing.its brill 6 years on totally solid.4.2g
I found out about this CPU yesterday on the bottom of AMD Virtual super resolution page. At launch it was 1059$USD(MSRP). The Google demons then showed me your video mouahahah
Chris these results are indeed very interesting, unfortunately now (especially in europe, with what is happening: we could be without gas and electricity in the winter) power consuption is not irrelevant anymore, like at all, so... do you know how much does this overclocked 5960x consume?
I7 5960x at 4.5 ghz consumes close to 500 watts. I know because I have one.
And yet I have it running on a 750w PSU with an RTX 3090... You need to lower your voltage man
@@TheGoodOldGamer Le I7 5960X même OC à 4,6 Ghz n'est absolument pas adapté à une RTX 3090. Beaucoup de watts gaspillé pour pas grand chose. Zen 2 fait tout aussi bien, avec beaucoup moins de consommation. Bref zen 2 est bien mieux au final.
@@TheGoodOldGamer It's all about tuning!
@@daddyalpha2648 Really ?!. I have one, at the same speed, with 32 GB 3200/T1, raid 1 of 1 TB 4 HD. 1 M2, 2 SDD, 2 HDD each of 2TB, and an RTX 3080 12GB underclocked, but that still reach 324 Watts in some games. Power supply is an corsair 1000w, and 2 monitores, plus VR headset. AH, and more USB devices,maybe 5. not even 300w the CPU consume at that speed.
Not surprised, ringbus and lots of cache.
How about performance per watt and combined platform (CPU, motherboard, ram, PSU) performance per dollar? (For each of these chips)
I guess I will have to test some of those games with my setup, looks like I choose right for a long lived platform. X99 FTW!
i"9" 6950X, golden sample, OC upto 4.6Ghz.
(14nm, 10C/20T, Broadwell-E)
Rampage V Edition 10, LGA 2011-3.
(OC socket with 2040 pins)
Corsair DP SE Blackout, 32GB, DDR4 3200 CL14.
(OC to 3600, Samsung B-Die)
RTX 3080 Ti ROG Strix OC 12GB.
I play at 4K, so the CPU would not be much of a problem anyways. Ha. 😃
Yeah at 4K anything on the list would work ;)
@@TheGoodOldGamer Once you go 4K 120Hz OLED there is no going back! 👀
@@MichaelJohnson-sb5si That's Ultimately the point I'm trying to get across. Even these OLD/Cheap CPUs can handle 120/144FPS. The brainwashing is STRONG out there right now.
@@TheGoodOldGamer What brainwashing are you referring to exactly?
@@davidandrew6855 The MOAR Cores and the You NEED Newest CPU for Gaming argument. It simply doesn't hold up.
Would I be able to oc the 5960x on a Chinese machinist x99?
For a brief moment i was thinking "damn finally one video without asking for money in every shape and form" and then BOOM he was asking for money LOL
Well this is his job, making videos for us to watch. I'm sure AdSense doesn't pay as much as you think, and nearly every tech channel has a community setup to join.
Yes, you missed something... The GPU side of things. When Zen1 came out the best gaming GPU was GTX 1080.
There’s always low res testing and there was a 1080ti which is about as fast as a 3060
@@TheGoodOldGamer But even then most reviews just used stock CPU which benefit Ryzen more and Intel less. Zen1 was definitely slow for gaming. Also keep in mind after 5960x there is likely only 1 gen ipc with Skylake until 12th gen came out so if you had 5960x, you could ride the intel wave for 6 gen before you got a double gen increase.
$1060 SRP Testing at 720p or DLSS upscaled some perf mode is not legit plus using nVid does not allow SAM. Next a $1700 8950x vs less than $499 1800x, at 720p and DLSS upscale np SAM Compare vs a Threadrpper
What's better, this haswell setup with the 5960X or the i3 12100?
No question about it, the 12100. It is new, has modern feature set, and a valid upgrade path with Rapture lake.
What are opinions on the new x99 mb on amazon and eBay
Your 5800x 4000cl16.. was it running fclk2000? Cuz if not, then this comparison is not valid. Run the 5800x with 3800mhz ram and post again
2000 FCLK Feel Free To Watch That Video