3:31 I'm glad you actually brought up CPU-Z because Linus didn't do it in his vid and I've been wondering what CPU-Z would say it is. It's interesting that it's recognised as a Xeon E5 and that it does indeed say the TDP is 250W! Also, in the comments on Linus' video someone mentioned that it is most likely a CPU microcode problem. As Intel never actually released this CPU, they also never released the BIOS microcode, which could lead to the absolute headache when trying to get it to work.
The part looks like its made of a 10c die (medium sized) due to 25MB L3 cache (thanks CPU-Z!) - same as is used in the E5-2687WV2 (8c/16t at 150w, with 4.0GHz turbo). The Core series top model (i7-4960X has only 15MB and 6c/12t which is the max out of the LCC - low core count die).
I strongly believe that this cpu has disabled cores to run 4.6 ghz for better single core performance, which is expected thing to get. Afaik the socket 2011 could run xeon cpus, like last haswell 1150 generation did.
@@MokokoMokoko I'm running a Haswell Xeon in my X99 board (S2011-3), and that mobo purchase was influenced by people running Xeons in S2011 boards, so... yes. Except saying "Haswell 1150" is kinda wrong since S1150 was the *desktop* Haswell socket. Just say Haswell.
@@samiraperi467 there were xeons for both haswell and haswell-e boards. The 1150 socket only got e3's which are pretty unexciting and only really good for low power ecc workloads.
@@samiraperi467 he was referring to the fact that the S1150 xeon E3-12xx V3 can be used in a desktop board as well (but only few models got a GPU enabled). The later models (ignoring broadwell hole) are the 6th gen xeons E3-12xx V5 skylake or 7th gen E3-12xx V6 kabylake are not working in the desktop 100/200 series boards and you need a Cxxx entry level server board (and same goes for the Xeon E-21xx/22xx - which are the 8th/9th gen Core series dies) Even that everything is the same die.. the fusing and bios code prevents cross-operation. Which is shame.
Please take it to the Intel labs and ask the lads there to show you how they debug such thing over the XDP plug. Would they be able to identify the fault causes (eg by internal counters), or is that exact cpu unit / die just faulty (or cursed) ?
agree, the variety of errors would seem to be a seating problem but with other processors working fine and the ops acting up on two boards + variety of BSODS + boot error variety + inaccurate temps...internal fault somewhere in the cpu or package connections.
@@sirmonkey1985 I would understand why Intel would want the CPU's securely destroyed, but would the Stock Exchange care? I mean it is still based on the same basic CPU Architecture as the retail platforms, so the majority of the CPU is still as flawed as any retail CPU would be. Things like this hardware are often securely destroyed just for peace of mind, of course what should happen to them VS what does happen to them are two different things, unfortunately for Intel, but, fortunately for us as the public.
I really appreciate videos like this, it shows even people like der8auer have troubleshooting situations that they cant figure out. The next time I find myself in a situation I cant seem to fix I wont be so hard on myself.
Wow, is this the CPU that EVGA SR-X was meant to get?! Some say Black Ops were special binned CPU's for mainframes. This was actually the second gen hexacores from Intel, the Gulftown Nehalems were first. I still daily an i7 990X Aurora R1.
Oh?I never knew they actually had a production run for it. I always thought the Black Ops cpu that exists in the wild are from engineering/testing samples or even a scam.
No this is not the CPU the EVGA SR-X was meant to get. This CPU was used in the stock exchange computers, where even a fraction of a second counted. Perhaps it will work properly if you get a motherboard that was used in the stock exchange computers and not common boards available for retail.
@@First-Name_Last-Name black ops CPU's were intended for the financial market because the programs they were using for stock trading need high clock speeds. They still produce black ops CPU's but they are special orders that will never see store shelves. Edit woops didn't see teksers post, lol.
@@LawrenceTimme Man how can you say it runs on LGA 1366 when he used an X79 board which is socket/LGA 2011? Everest also had it's share of black ops CPU's but this ain't it.
Worth watching, always curious about these black ops units, but I'm not about to buy a setup to test it myself as the chance of ending up with something useful is so low, so I appreciate you going for it, and I will always support your trying to make it work by watching and liking your videos, and throwing the occasional; comment to the algorithm.
If you can find the original server chassis they were in, it works fine in those. When I bought mine the guy sent me screenshot from the server and it was running good. Unfortunately I cannot afford the shipping of the whole server and just bought the CPU.
All you got was proof he managed to boot it. It seems like electromigration like what you would see in a CPU from this generation that spent years on a 24/7 high voltage overclock
I was told these were in a Dell Poweredge Chasis w/ special liquid cooling configuration. I spoke with a recycle who was scrapping them and that's how I got mine. He had already scrapped all the water blocks and servers, I asked for a complete one but no dice.
Go into the DIGI+ settings and max out all the power related settings. This will likely allow the CPU to deal with the higher draw. This is needed for the E5-1680v2 also.
Love stuff like this with exotic,Rare or ES components i just find it fascinating to see them been tried out see what does and what doesn't not work and how buggy they were etc. :)
I had an ES slot 1 pentium III in my old system and I didnt know until years later, bought an ES xeon 2011v3 cpu that was very different from the final product, it had way lower voltage and very low core frequency, Definitely not as good as the final product but still cool, strange cpu, 12c 24t 1.5ghz 2.4boost.
Spent some time on google, the info on the English sites is from a chines article titled '【有趣】第21期：极其罕见、极其强大、极其有趣的英特尔处理器' , found a few mentions on chines server news sites. Some mentions of Dell systems running CenOS using the CPU, may be info on some Linux site from devices phoning home or reporting bugs? There are screenshots of the BIOS on dell mobos, on the dell download site that BIOS version is skipped.
If you happen to still have the hardware it might be interesting to see how the quad and/or hex core CPUs scale over the generations. It's pretty interesting to see those results and where things stand. Even if you weren't able to get the primary focus going it still had information enough to be interesting.
Yes and how does it compare to the i7 970 socket 1366 X58, I remember that thing being very efficient compared to my i7 920, and it had 6cores + HT. That was around 2010. It worked fine on 4GHz and did not get very hot either.
I'd actually get a Xeon E5 v1/v2 generation motherboard with the C602 chipset as this is what the chip was designed for (avoid C606 due to SAS issues). The official maximum speed on the memory controller is 1866 Mhz These chips were sold to high frequency traders which needed the best single threaded performance. A board with ECC would also be helpful. The use-case explains why HyperThreading and Turbo was disabled as the trading algorithms needed predictable latency and those technologies introduce too much jitter for processing. This isn't the first time Intel made a chip that was specialized for this market: look up the Xeon X5698 as that was a dual core 4.4 Ghz model based on the Westmere architecture.
Your English is very good, your frequent use of the word "Like" is almost up to the level as most younger English speakers in the USA. Havent noticed in your other videos before, really enjoy the content you produce.
I have found a lot of joy in getting older hardware and overclocking to squeeze max performance. I have an AMD 955 Black Edition overclocked to 3.8GHz(going to get a liquid cooler to push it further0, Intel i5-6600K overclocked to 4.8GHz, and an Intel i7-7700K overclocked to 4.8GHz(getting liquid cooler for this one too since it is just limited by heat). Enjoyed the video! :)
BlackOps has strange temperature walls that allow the frequency to run in maximum mode when the temperature is below 20 degrees Celsius, the performance will be 10 ~15% lower during 40 degrees Celsius (but above 20), and 15 ~20% lower when temperature is higher than 40 degrees Celsius
Hey, I just noticed your Stargate tattoo. That seems like the one tattoo idea that I wouldn't regret in the future. That's what's really kept me from getting a tattoo - the thought I would regret it later in life.
Current + high temp is deadly for any CPU, however I found Intels to be quite fragile compared to AMD chips in regards of lifespan after heavy OC. Prime95 or continuous hammering in Cinebench (or any other load, especially AVX) will eat an overclocked CPU quickly. I suspect the Black Ops was calculated for a certain lifespan after which it simply got replaced before problems would arise, or they ran under (chilled) water.
@der8auer : Roman what exactly is an intel "Black Ops" CPU supposed to be ? Is it like a technology demonstrator, engineering sample, early alpha/beta version or something further along the pipeline ?
i have a x3470 and r9270x would upgrading the gpu help for war thunder or should i replace the whole computer pc parts are really expensive and i have a dirtbike addiction to feed😅
Hey, loved the video as always! i have a R7 5800x with the exact same problem, instant crashes, 99c, auto reads of 1.5v, 750w total tdp at ryzenmaster, a lot off bios features missing, my applyed voltages didnt matter.. it was only 1.5 or 1.1v fixed. . We tested with another mb and same results, with another processor it was fine just as you said. i had to fix 4.2ghz and the voltage automatically drops to 1.1fixed. lost 6% points on r23. i was stable in prime95 for 5 hours with temps arround 50c max, suddnly it spikes to 99 and crashes, but only in this cenario. gaming we ran tests for 24h and no crashes or temps spikes. P.s. windows defender runs appears to spike the cores to 90c.
A bit of a shot in the dark that might make me sound like an idiot but what if this CPU ran the stock 1.55V for multiple years and has degraded from that ? Would need to be quite serious to even refuse to work with slowmode but idk
Im guessing electromigration from this is exactly the issue. This CPU was made for a good time not a long time, I'd guess in an environment with hot-swap, redundant fault tolerant processing and hot spares, where ther'e a support contract and the machine cost so much to buy and run it didnt really matter if you had to replace the CPUs every few years (ie mainframes). If it could have run like that for 10y they would have just sold them
I thought for sure you were going to pull out the 8 core Xeon 1680v2. That's by far the best CPU available from this era. I ran one at 4.6GHz up until about 6 months ago. I don't think they're hard to get these days but if I send you one would you be willing to do another video showing what X79 is really capable of? I also have what I think is one of the best 4820K samples but I'm not sure if I'm willing to part with that.
How did you overclock that Xeon ? I'm currently using a W-2145, 8 core myself in an HP Z4 G4. It breezes past the 3DMark tests at very low temperatures.
@@grahamstevenson1740 It was one of the last unlocked Xeon models released at the time. A lot of them came out of the 2013 iMac pro and were very expensive. 8 Ivy-E cores with 25mb of cache gets hot compared to its default speed but it's a beast of chip that's basically 2700X performance in many cases.
very interesting, It looks like it's more a workstation processor than anything else. If that's the case then it may work in a workstation/server motherboard from the time, I think those chipsets were C602, C604, C606, C608. My guess is it's unstable due to requesting a power state that's not supported on consumer boards. I'd also be very curios to find out if ECC works with it or not.
my guess is that the hi voltage is making the cpu act up. i would try to manually set it with constant voltage settings of similiar cpu (22nm ivy bridge -ep/ex whatever) to see what it does. Back in amd fx era, i experienced many times weirdness when the cpu was high-ish temps but not hothot, like even 70c would cause random crashes and whatnot
It might have something to do with the fact that no consumer motherboard exists for this CPU. They probably had some that work with it in a lab somewhere, but... HAD!
Although not I currently using Windows and Linux as my Dail driver for 2+ yrs, I still have some memory for tweaking around LGA15xx-2011 sockets ES CPUs. This chip is possibly a very early test sample for architecture-thermal verification (than common Intel confidential ES samples, one can verify it with HWiNFO for more info on stepping and prevention no. etc.)
That's indeed high voltage, my 4930k @ 4.7Ghz is at 1.4v (but that's a good sample that does 6Ghz on LN2), and another one that does 4.6Ghz at 1.4v as well. Good ol Rampage 4 board (got 2 red ones and 1 black edition 😁), remember when those top-of-the-line boards didn't cost an arm and leg. I'm still using this system today (yep, I'm looking to upgrade, but availability is horrendous right now). I do like it when some reviews come back to that old hardware and compare them to today's mainstream ones. 👍
I remember when I cried having to pay like 550 or 600 euro's for an X299 DARK way back.....last year I had to pay 1250 for a Z790 Extreme, damn that's not even HEDT...
My 12700K for stock perfomance, has 1.42 volts for the 5GHz turbo. Now with partial OC and undervolt it stays between 1.23 volts or 1.3. I think my MSI motherboard is broken (i haven't updated the BIOS).
Looks like this thing was an experiment about a potential high performance server CPU, and that it was ditched because it was highly unstable/some internals were fried during the experiments. Maybe the HT is inactive cause the scheduler was damaged ? Maybe it has a lot of erosion ? maybe some internal nodes are dead ?
I would be curious ti see if this is running a bit better on a server board with ecc ram, as it's being referenced as a Xeon, it might help it a bit 🤷♂
The issue with normal boot up process is in ME region of bios. It must be downgraded from 8.x to 7.x and only 5MB (AMT) region is capable of driving this beast. Therefore simply replacing ME-region in bios-dump is not the way to get it working: boot block of IV rampage must be rewritten and microcode flashed from server boards of Sandy Bridge-EP era. I did managed to get BO cpu working, but it was 8yrs ago. And the first barely positive try was in lga2011 supermicro dual cpu board.
@@romanm.4763 idk, but on desktop board it was the only solution. Might be it's related to the first bunch of ivy bridge skus, that worked properly with exact microcode and old ME.
Have you tried overvolting a bit? Seems like the core is very unstable and only stable at 1.5v at low clock. Or don't use uefi, rather use legacy bios?
I'm quite familiar with this platform (as up until recently I was still using a heavily overclocked X79 based system as my main PC) and yeah, that voltage is stupidly high (my older 3930k only needed 1.35v to hit 4.4ghz) so I'm I wondering if the motherboards can even reliably supply 1.55v (especially when they're around 10 years old at this point). The other thing that jumps out as a potential issue is the ram you've chosen (atleast on the older sandy bridge-e CPUs 2666mhz was a hell of a lot to ask, around 1866mhz is more reasonable and you've got plenty of bandwidth with quad channel anyway) I doubt it's the main problem though as it looks like you tried running with default 1333mhz profile.
Yeah the voltage and the ripple, because those MBs are old, ie, how are the caps. Yeah the RAM, ie, too high a Hz and the CPU timings might not be able to keep up.
@@SidneyCritic and the last thing (which I was tempted to bring up) is just basic crap like "the thermal paste on the mobo chipset is probably hard as a rock by now". That x79 I recently upgraded from became my wife's gaming system at the same time, and while I was swapping the bits around I decided to check out the thermal paste under the chipset cooler, and yeah, I suddenly became aware of why that computer had been crashing on me so much. About the 10 year mark you definitely have to check on things like that (and cmos batteries).
@@fuzz11111111 I have to run a Z87 because it has PCI and parallel port for old hardware. I was thinking of upgrading recently, but then I realised I had to throw away good, but old, hardware and struggle with Win10 to get old DOS engineering stuff working. I currently component-level repair 3rd and 4th gen motherboards for fun - lol -.
I would guess neither BIOS nor OS will have a valid microcode to upload on this CPU, so at best will try something "close" but mismatching, so it will not initialize and function properly.
@@stanimir4197 They don't submit blobs for one-off prototypes (e.g. ones made to play with some new ideas which end to be dead-ends). Edit: to avoid speculation it would be relatively easy to see at linux boot log and if it indeed managed to recognize CPU as something unique and also had correct microcode to upload for it.
@@ridingtheflow6752 A 1 off prototype would be an ES and be unlikely to be clocked that high. I'd say this CPU was made for a good time not a long time and if they could handle a decade under that voltage they would have been sold retail that way. Plenty of environments where a CPU dying would be a tiny part of the maintenance contract cost and any unreliability would be quickly picked up by redundant processing.
This whole thing kinda look like my experience to mine when i moved for college. 8 hours of trouble shooting sudenlly work, then unplug the hdmi cable then it stop working
Could the lack of hyperthreading be due to Windows’ spectre/meltdown mitigations, or was that actually how this CPU was made? Try booting Linux again, with mitigations disabled.
My old 4930k was far from gold sample and I did 4.8 GHz easily with 1.4 vcore, could even use medium LLC for lower temps if HT was off. I was hoping to see past 5 GHz on a 250w cpu hahaha
Thanks for reminding me that I need to replace my 3930k machine, but man, I can do 1.5v on a 240mm AIO cooler and hit 4.5Ghz, but man does it draw a lot of power. The whole CPU reporting as a Xeon might make me think that some boards just aren't stable with Xeon chips, particularly at these clock speeds.
Should try different types of ram, ive had some CPUs that just WOULD NOT work with some types of ram, even tho they were both technically compatible, and both perfectly fine when swapped with different ram/cpu
I have an old i7 3930k, which I changed some time ago for obvious reasons, but I can really say that it is a beast, in its time nothing really looked like it, until the first commercial 6 core intel processors came out, it was extremely powerful. PS: I still have it and I have an Asus Rampage IV Gene with 2x8 Corsair Vengeance RAM.
I guess if it's pre-production it might want all the old stuff, ie, RAM and MB, or what was around at the time. RAM to CPU is pretty critical. I had something like that decades ago when I tried to run a mobile processor in a PC motherboard - I didn't know mobile CPUs were different because it looked the same as a normal CPU - the "M" after the # was the clue lol -, so the guy in the shop tried 5 of the same MB until one worked reliability - the guy didn't know it was a mobile CPU either -.
Man I cant stop laughing, this is my exact experience last the last week, trying to get an old x99 system up and at a decent overclock. I feel your pain. Been so rough that I started looking into very in depth overclocking to try to learn something that might help.
Yeah, I ended up in the same boat after my 7 year old 5960x @4.8ghz died a week ago. Tried to replace it with an eBay buy and got stuck with a CPU with a dead PCIE controller.
What about reverting back to a very early bios? I've had to replace the bios chip to get an older version of the bios to run. At least on these old boards you can actually get a bios chip that is easily replaceable
HT would be turned off or disabled for HFT platforms you need consistent fast computeing without possible different logic latency paths you would run two or more parallel code stacks and has race to win for for zero loss and the mtb would be either an Intel, supermicro, Dell. Fujitsu, oracle , HP, blades, could even been a Cisco device. IBM not some consumer retail boards. Get a server mtb, with no gpu just an on board vga
Seems like the CPU is toast. Electro migration from high voltage has taken another Processor. Thanks for trying with this limited edition proc Der8auer.
Unlocked e5 2667v2 or e5 2680 v2 with disabled cores ? And no voltage limits she aint built for a longtime shes built for a good time kinda like a cuvic with an ebay turbo
Now this video gives me headaches of wanting some LGA2011 action...sure, I can get the CPUs from China or somewhere but finding a board will be another problem. Even if they'd run hot it will still be faster than the Core 2 Duo T7200 in a laptop I have...also running XP for older programs, newer boards doesn't have drivers for that
I'd say if it could run at 1.5v for 10 years they would have just sold them like that. Guessing the reason they werent retail is running them that way isnt compatible with a retail warranty.
![[UNCUT] รัฐสั่งหยุดปล่อยคลื่น! แต่หยุดพลังปะทุ “น้องหญิง” ไม่ได้ I คนดังนั่งเคลียร์](/img/n.gif)
3:31 I'm glad you actually brought up CPU-Z because Linus didn't do it in his vid and I've been wondering what CPU-Z would say it is. It's interesting that it's recognised as a Xeon E5 and that it does indeed say the TDP is 250W!
Also, in the comments on Linus' video someone mentioned that it is most likely a CPU microcode problem. As Intel never actually released this CPU, they also never released the BIOS microcode, which could lead to the absolute headache when trying to get it to work.
The part looks like its made of a 10c die (medium sized) due to 25MB L3 cache (thanks CPU-Z!) - same as is used in the E5-2687WV2 (8c/16t at 150w, with 4.0GHz turbo). The Core series top model (i7-4960X has only 15MB and 6c/12t which is the max out of the LCC - low core count die).
I strongly believe that this cpu has disabled cores to run 4.6 ghz for better single core performance, which is expected thing to get. Afaik the socket 2011 could run xeon cpus, like last haswell 1150 generation did.
@@MokokoMokoko I'm running a Haswell Xeon in my X99 board (S2011-3), and that mobo purchase was influenced by people running Xeons in S2011 boards, so... yes. Except saying "Haswell 1150" is kinda wrong since S1150 was the *desktop* Haswell socket. Just say Haswell.
@@samiraperi467 there were xeons for both haswell and haswell-e boards. The 1150 socket only got e3's which are pretty unexciting and only really good for low power ecc workloads.
@@samiraperi467 he was referring to the fact that the S1150 xeon E3-12xx V3 can be used in a desktop board as well (but only few models got a GPU enabled). The later models (ignoring broadwell hole) are the 6th gen xeons E3-12xx V5 skylake or 7th gen E3-12xx V6 kabylake are not working in the desktop 100/200 series boards and you need a Cxxx entry level server board (and same goes for the Xeon E-21xx/22xx - which are the 8th/9th gen Core series dies) Even that everything is the same die.. the fusing and bios code prevents cross-operation. Which is shame.
Please take it to the Intel labs and ask the lads there to show you how they debug such thing over the XDP plug. Would they be able to identify the fault causes (eg by internal counters), or is that exact cpu unit / die just faulty (or cursed) ?
agree, the variety of errors would seem to be a seating problem but with other processors working fine and the ops acting up on two boards + variety of BSODS + boot error variety + inaccurate temps...internal fault somewhere in the cpu or package connections.
@@Joe-Dead I just needs the CPU micro code and to be run colder.
If he did they'd confiscate it. These CPU's are supposed to be securely destroyed after they're decommissioned.
@@sirmonkey1985 unlikely
@@sirmonkey1985 I would understand why Intel would want the CPU's securely destroyed, but would the Stock Exchange care?
I mean it is still based on the same basic CPU Architecture as the retail platforms, so the majority of the CPU is still as flawed as any retail CPU would be.
Things like this hardware are often securely destroyed just for peace of mind, of course what should happen to them VS what does happen to them are two different things, unfortunately for Intel, but, fortunately for us as the public.
I really appreciate videos like this, it shows even people like der8auer have troubleshooting situations that they cant figure out.
The next time I find myself in a situation I cant seem to fix I wont be so hard on myself.
Wow, is this the CPU that EVGA SR-X was meant to get?!
Some say Black Ops were special binned CPU's for mainframes. This was actually the second gen hexacores from Intel, the Gulftown Nehalems were first.
I still daily an i7 990X Aurora R1.
Oh?I never knew they actually had a production run for it.
I always thought the Black Ops cpu that exists in the wild are from engineering/testing samples or even a scam.
No this is not the CPU the EVGA SR-X was meant to get. This CPU was used in the stock exchange computers, where even a fraction of a second counted.
Perhaps it will work properly if you get a motherboard that was used in the stock exchange computers and not common boards available for retail.
@@First-Name_Last-Name black ops CPU's were intended for the financial market because the programs they were using for stock trading need high clock speeds. They still produce black ops CPU's but they are special orders that will never see store shelves.
Edit woops didn't see teksers post, lol.
This is an older version that runs at 4.4ghz called Everest for lga1366
@@LawrenceTimme Man how can you say it runs on LGA 1366 when he used an X79 board which is socket/LGA 2011? Everest also had it's share of black ops CPU's but this ain't it.
I like this kind of videos. It's interesting to see what weird stuff there is around.
Worth watching, always curious about these black ops units, but I'm not about to buy a setup to test it myself as the chance of ending up with something useful is so low, so I appreciate you going for it, and I will always support your trying to make it work by watching and liking your videos, and throwing the occasional; comment to the algorithm.
If you can find the original server chassis they were in, it works fine in those. When I bought mine the guy sent me screenshot from the server and it was running good. Unfortunately I cannot afford the shipping of the whole server and just bought the CPU.
All you got was proof he managed to boot it. It seems like electromigration like what you would see in a CPU from this generation that spent years on a 24/7 high voltage overclock
Can you at least tell us who made the server (or a model number)? Really curious.
I was told these were in a Dell Poweredge Chasis w/ special liquid cooling configuration. I spoke with a recycle who was scrapping them and that's how I got mine. He had already scrapped all the water blocks and servers, I asked for a complete one but no dice.
Thank you for making the video, I appreciate the efforts!
Go into the DIGI+ settings and max out all the power related settings. This will likely allow the CPU to deal with the higher draw. This is needed for the E5-1680v2 also.
Omg I’ve been looking for a review about this exact cpu since ages ! Thanks !
Thanks for showing us. Is little frustrating see everything on YT go as smooth as planned.
Love stuff like this with exotic,Rare or ES components i just find it fascinating to see them been tried out see what does and what doesn't not work and how buggy they were etc. :)
I had an ES slot 1 pentium III in my old system and I didnt know until years later, bought an ES xeon 2011v3 cpu that was very different from the final product, it had way lower voltage and very low core frequency, Definitely not as good as the final product but still cool, strange cpu, 12c 24t 1.5ghz 2.4boost.
@@IvanOoze1990 Thats bizarre and super cool at the same time,Definitely would have kept in a collection :)
Love videos like this. thx for making them :)
Lawrence Timme has a video running the CPU, mentions you need to keep the CPU under 20C to run.
Spent some time on google, the info on the English sites is from a chines article titled '【有趣】第21期：极其罕见、极其强大、极其有趣的英特尔处理器' , found a few mentions on chines server news sites. Some mentions of Dell systems running CenOS using the CPU, may be info on some Linux site from devices phoning home or reporting bugs?
There are screenshots of the BIOS on dell mobos, on the dell download site that BIOS version is skipped.
If you happen to still have the hardware it might be interesting to see how the quad and/or hex core CPUs scale over the generations. It's pretty interesting to see those results and where things stand. Even if you weren't able to get the primary focus going it still had information enough to be interesting.
I would love to see a comparison with a 4930K.
Yes and how does it compare to the i7 970 socket 1366 X58, I remember that thing being very efficient compared to my i7 920, and it had 6cores + HT. That was around 2010. It worked fine on 4GHz and did not get very hot either.
I'd actually get a Xeon E5 v1/v2 generation motherboard with the C602 chipset as this is what the chip was designed for (avoid C606 due to SAS issues). The official maximum speed on the memory controller is 1866 Mhz These chips were sold to high frequency traders which needed the best single threaded performance. A board with ECC would also be helpful. The use-case explains why HyperThreading and Turbo was disabled as the trading algorithms needed predictable latency and those technologies introduce too much jitter for processing.
This isn't the first time Intel made a chip that was specialized for this market: look up the Xeon X5698 as that was a dual core 4.4 Ghz model based on the Westmere architecture.
Your English is very good, your frequent use of the word "Like" is almost up to the level as most younger English speakers in the USA.
Havent noticed in your other videos before, really enjoy the content you produce.
Great video mate, I can't imagine how frustrating it would be to do this video & then have to do it all over in english.
Thank you for this video 👌
I have found a lot of joy in getting older hardware and overclocking to squeeze max performance. I have an AMD 955 Black Edition overclocked to 3.8GHz(going to get a liquid cooler to push it further0, Intel i5-6600K overclocked to 4.8GHz, and an Intel i7-7700K overclocked to 4.8GHz(getting liquid cooler for this one too since it is just limited by heat). Enjoyed the video! :)
I did enjoy this video in spite of the results. You have great patience!
Wow didnt expect to see a new video about an ivybridge-e chip any time soon lol, sweet
BlackOps has strange temperature walls that allow the frequency to run in maximum mode when the temperature is below 20 degrees Celsius, the performance will be 10 ~15% lower during 40 degrees Celsius (but above 20), and 15 ~20% lower when temperature is higher than 40 degrees Celsius
love this kind of video, even if you didn't manage to do everything you wanted :)
I love those 7-segment debug displays.
9:40 Mixed graphs with power is so valuable!
@11:20 I'd try to separate RAM clock? If BIOS allows that?
Hey, I just noticed your Stargate tattoo. That seems like the one tattoo idea that I wouldn't regret in the future.
That's what's really kept me from getting a tattoo - the thought I would regret it later in life.
My experience with ivy bridge is that,
They degrade quite easily with an overclock over time
Current + high temp is deadly for any CPU, however I found Intels to be quite fragile compared to AMD chips in regards of lifespan after heavy OC.
Prime95 or continuous hammering in Cinebench (or any other load, especially AVX) will eat an overclocked CPU quickly. I suspect the Black Ops was calculated for a certain lifespan after which it simply got replaced before problems would arise, or they ran under (chilled) water.
Thanks for this informative video
do you think that chip was designed for a special 3 channel memory configuration? what board can run three channels?
@der8auer : Roman what exactly is an intel "Black Ops" CPU supposed to be ? Is it like a technology demonstrator, engineering sample, early alpha/beta version or something further along the pipeline ?
Think those were customs for high speed stock trading
Thanks for the video.
Was wondering if you gave it 1.65 to 1.7 volts would that help with stability?
higher voltage would literally cook the cpu.
I was relieved after confirming that there were areas that der8auer could not do because he was also human.
i have a x3470 and r9270x would upgrading the gpu help for war thunder or should i replace the whole computer pc parts are really expensive and i have a dirtbike addiction to feed😅
Hey, loved the video as always! i have a R7 5800x with the exact same problem, instant crashes, 99c, auto reads of 1.5v, 750w total tdp at ryzenmaster, a lot off bios features missing, my applyed voltages didnt matter.. it was only 1.5 or 1.1v fixed. . We tested with another mb and same results, with another processor it was fine just as you said. i had to fix 4.2ghz and the voltage automatically drops to 1.1fixed. lost 6% points on r23. i was stable in prime95 for 5 hours with temps arround 50c max, suddnly it spikes to 99 and crashes, but only in this cenario. gaming we ran tests for 24h and no crashes or temps spikes. P.s. windows defender runs appears to spike the cores to 90c.
From what i remember, these were dedicated LN2 CPU's.
In the dell server they pulled from they ran on chilled water.
This. You need this dell sever.
A bit of a shot in the dark that might make me sound like an idiot but what if this CPU ran the stock 1.55V for multiple years and has degraded from that ?
Would need to be quite serious to even refuse to work with slowmode but idk
Could be, but the way we can completely compensate the degradation is to raise voltage or downclock. He downclocked to 2.8 and nothing changed
@@ramasso ...Kernel panic and the like - most likely it's the memory controller that's gone.
@@ramasso He even downclocked to 800MHz with slowmode yeah
Im guessing electromigration from this is exactly the issue. This CPU was made for a good time not a long time, I'd guess in an environment with hot-swap, redundant fault tolerant processing and hot spares, where ther'e a support contract and the machine cost so much to buy and run it didnt really matter if you had to replace the CPUs every few years (ie mainframes).
If it could have run like that for 10y they would have just sold them
Maybe try the ASRock extreme boards, they are compatible with a shockingly large amounts of CPUs.
I thought for sure you were going to pull out the 8 core Xeon 1680v2. That's by far the best CPU available from this era. I ran one at 4.6GHz up until about 6 months ago. I don't think they're hard to get these days but if I send you one would you be willing to do another video showing what X79 is really capable of?
I also have what I think is one of the best 4820K samples but I'm not sure if I'm willing to part with that.
How did you overclock that Xeon ? I'm currently using a W-2145, 8 core myself in an HP Z4 G4. It breezes past the 3DMark tests at very low temperatures.
@@grahamstevenson1740 It was one of the last unlocked Xeon models released at the time. A lot of them came out of the 2013 iMac pro and were very expensive. 8 Ivy-E cores with 25mb of cache gets hot compared to its default speed but it's a beast of chip that's basically 2700X performance in many cases.
"Persistence" must be your middle name. You went far and beyond what I'd have done with that odd-ball CPU.
BlackOps CPUs only work when in full tactical gear and balaclavas.
very interesting, It looks like it's more a workstation processor than anything else. If that's the case then it may work in a workstation/server motherboard from the time, I think those chipsets were C602, C604, C606, C608. My guess is it's unstable due to requesting a power state that's not supported on consumer boards. I'd also be very curios to find out if ECC works with it or not.
my guess is that the hi voltage is making the cpu act up.
i would try to manually set it with constant voltage settings of similiar cpu (22nm ivy bridge -ep/ex whatever) to see what it does.
Back in amd fx era, i experienced many times weirdness when the cpu was high-ish temps but not hothot, like even 70c would cause random crashes and whatnot
Too bad couldnt get the cpu to run stable but still enjoyed the video. Also did you ever try changing the VTT voltage to 1.150 from the stock 1.050?
I would try 1.25v VTT and 2nd VTT 1.24926v on the RIVE-BE
This seems to be the CPU equivalent to a screamapillar.
Actually the elpidas on the kingston t1s are able to hit around 2600c7, which is insane in my opinion and the best memory ever created.
Even in Quad channel?
@@NumbDiggers1998 Yup, thats when they shine
It might have something to do with the fact that no consumer motherboard exists for this CPU. They probably had some that work with it in a lab somewhere, but... HAD!
I have a e5 1680 v2 under an open loop 480mm it's cranking at 4.8. I don't remember the voltage but it was very close to the safety limit.
Although not I currently using Windows and Linux as my Dail driver for 2+ yrs, I still have some memory for tweaking around LGA15xx-2011 sockets ES CPUs. This chip is possibly a very early test sample for architecture-thermal verification (than common Intel confidential ES samples, one can verify it with HWiNFO for more info on stepping and prevention no. etc.)
That's indeed high voltage, my 4930k @ 4.7Ghz is at 1.4v (but that's a good sample that does 6Ghz on LN2), and another one that does 4.6Ghz at 1.4v as well.
Good ol Rampage 4 board (got 2 red ones and 1 black edition 😁), remember when those top-of-the-line boards didn't cost an arm and leg. I'm still using this system today (yep, I'm looking to upgrade, but availability is horrendous right now).
I do like it when some reviews come back to that old hardware and compare them to today's mainstream ones. 👍
I remember when I cried having to pay like 550 or 600 euro's for an X299 DARK way back.....last year I had to pay 1250 for a Z790 Extreme, damn that's not even HEDT...
My 12700K for stock perfomance, has 1.42 volts for the 5GHz turbo.
Now with partial OC and undervolt it stays between 1.23 volts or 1.3.
I think my MSI motherboard is broken (i haven't updated the BIOS).
Looks like this thing was an experiment about a potential high performance server CPU, and that it was ditched because it was highly unstable/some internals were fried during the experiments. Maybe the HT is inactive cause the scheduler was damaged ? Maybe it has a lot of erosion ? maybe some internal nodes are dead ?
I have one of these! Awesome video.
Some of the comments have more info than his video here provides.
Would you consider taking a look at the other Everest SKUs? The QX9775, X5492, X5698
No voltage read from two memory channels definitely means that the memory controller is fried.
the fast STARTUPPPPPPPPPPPP!
I wonder if the memory controller on the CPU is not good enough to handle those ram speeds?
Aaaaaah i really Love these kind of stuff! Try to get one 995X!
I would be curious ti see if this is running a bit better on a server board with ecc ram, as it's being referenced as a Xeon, it might help it a bit 🤷♂
probably just a false label due to the lack of proper cpu microcodes in the bios.
The issue with normal boot up process is in ME region of bios. It must be downgraded from 8.x to 7.x and only 5MB (AMT) region is capable of driving this beast. Therefore simply replacing ME-region in bios-dump is not the way to get it working: boot block of IV rampage must be rewritten and microcode flashed from server boards of Sandy Bridge-EP era. I did managed to get BO cpu working, but it was 8yrs ago. And the first barely positive try was in lga2011 supermicro dual cpu board.
It's 22nm ivy bridge. Why does it need sandy bridge me ?
@@romanm.4763 idk, but on desktop board it was the only solution. Might be it's related to the first bunch of ivy bridge skus, that worked properly with exact microcode and old ME.
Have you tried overvolting a bit?
Seems like the core is very unstable and only stable at 1.5v at low clock.
Or don't use uefi, rather use legacy bios?
Tried all of that :D From very low clock and only 1.2 Volt up to 1.7 Volt in small steps. Didn't matter for the CPU
@@der8auer-en thats sad. Either way it is a special piece of decoration
I'm quite familiar with this platform (as up until recently I was still using a heavily overclocked X79 based system as my main PC) and yeah, that voltage is stupidly high (my older 3930k only needed 1.35v to hit 4.4ghz) so I'm I wondering if the motherboards can even reliably supply 1.55v (especially when they're around 10 years old at this point).
The other thing that jumps out as a potential issue is the ram you've chosen (atleast on the older sandy bridge-e CPUs 2666mhz was a hell of a lot to ask, around 1866mhz is more reasonable and you've got plenty of bandwidth with quad channel anyway) I doubt it's the main problem though as it looks like you tried running with default 1333mhz profile.
Yeah the voltage and the ripple, because those MBs are old, ie, how are the caps.
Yeah the RAM, ie, too high a Hz and the CPU timings might not be able to keep up.
@@SidneyCritic and the last thing (which I was tempted to bring up) is just basic crap like "the thermal paste on the mobo chipset is probably hard as a rock by now".
That x79 I recently upgraded from became my wife's gaming system at the same time, and while I was swapping the bits around I decided to check out the thermal paste under the chipset cooler, and yeah, I suddenly became aware of why that computer had been crashing on me so much.
About the 10 year mark you definitely have to check on things like that (and cmos batteries).
@@fuzz11111111 I have to run a Z87 because it has PCI and parallel port for old hardware. I was thinking of upgrading recently, but then I realised I had to throw away good, but old, hardware and struggle with Win10 to get old DOS engineering stuff working.
I currently component-level repair 3rd and 4th gen motherboards for fun - lol -.
Who knows what that CPU went through before you received it.
I would guess neither BIOS nor OS will have a valid microcode to upload on this CPU, so at best will try something "close" but mismatching, so it will not initialize and function properly.
It'd be exceptionally unlikely linux not to have a microcode set, given Intel is one of the largest code contributors to the kernel.
@@stanimir4197 They don't submit blobs for one-off prototypes (e.g. ones made to play with some new ideas which end to be dead-ends). Edit: to avoid speculation it would be relatively easy to see at linux boot log and if it indeed managed to recognize CPU as something unique and also had correct microcode to upload for it.
I thought so too. If those CPUs aren't meant to be sold, motherboards are unlikely to have correct microcode update.
@@ridingtheflow6752 A 1 off prototype would be an ES and be unlikely to be clocked that high. I'd say this CPU was made for a good time not a long time and if they could handle a decade under that voltage they would have been sold retail that way. Plenty of environments where a CPU dying would be a tiny part of the maintenance contract cost and any unreliability would be quickly picked up by redundant processing.
@@NikolayIslentev It most cases it's the OS that updates the microcode, so the BIOS/motherboard just need the most basic boot up to succeed.
This whole thing kinda look like my experience to mine when i moved for college. 8 hours of trouble shooting sudenlly work, then unplug the hdmi cable then it stop working
LOL that classic MSI marketing “Military Class” … they didn’t know the term is “miltary grade” 😂
I wonder if the OS bundled microcode was causing issues?
have you tried just to set base clock or multiplier 4,8/4,9 and onle single channel memory. don't change anything else.
Good illustration of electromigration and why lower voltage is better for your CPU long term.
I love the mboards that you just picked up on ebay.
Could the lack of hyperthreading be due to Windows’ spectre/meltdown mitigations, or was that actually how this CPU was made?
Try booting Linux again, with mitigations disabled.
No, bios shots at 3:38 showed "No Intel-HT"
Fun at ur expense
Luv it
haha :D
@@der8auer-en good show
My old 4930k was far from gold sample and I did 4.8 GHz easily with 1.4 vcore, could even use medium LLC for lower temps if HT was off. I was hoping to see past 5 GHz on a 250w cpu hahaha
Thanks for reminding me that I need to replace my 3930k machine, but man, I can do 1.5v on a 240mm AIO cooler and hit 4.5Ghz, but man does it draw a lot of power. The whole CPU reporting as a Xeon might make me think that some boards just aren't stable with Xeon chips, particularly at these clock speeds.
Should try different types of ram, ive had some CPUs that just WOULD NOT work with some types of ram, even tho they were both technically compatible, and both perfectly fine when swapped with different ram/cpu
I have an old i7 3930k, which I changed some time ago for obvious reasons, but I can really say that it is a beast, in its time nothing really looked like it, until the first commercial 6 core intel processors came out, it was extremely powerful.
PS: I still have it and I have an Asus Rampage IV Gene with 2x8 Corsair Vengeance RAM.
Tried changing to a 2x4gb RAM module config?
It might not like to have more than 8gb total on board if it was a mid-development chip.
Yes. Tried single, and dual dimms also in different configs and speed. 8GB and 4GB modules
I guess if it's pre-production it might want all the old stuff, ie, RAM and MB, or what was around at the time. RAM to CPU is pretty critical.
I had something like that decades ago when I tried to run a mobile processor in a PC motherboard - I didn't know mobile CPUs were different because it looked the same as a normal CPU - the "M" after the # was the clue lol -, so the guy in the shop tried 5 of the same MB until one worked reliability - the guy didn't know it was a mobile CPU either -.
Man I cant stop laughing, this is my exact experience last the last week, trying to get an old x99 system up and at a decent overclock. I feel your pain. Been so rough that I started looking into very in depth overclocking to try to learn something that might help.
Yeah, I ended up in the same boat after my 7 year old 5960x @4.8ghz died a week ago. Tried to replace it with an eBay buy and got stuck with a CPU with a dead PCIE controller.
What about reverting back to a very early bios? I've had to replace the bios chip to get an older version of the bios to run. At least on these old boards you can actually get a bios chip that is easily replaceable
The CPU is more than 10 years old (IvyBridge uarch) and therefore it's either a trash or museum exhibit, depending on which way you look at it.
i enjoy this type of content. everything is stale and boring in the tech world right now! this was refreshingly different
Seems a bug in the CPU, no firmware update available?
I had an Atom C2550 with a failing cache that had similar spurious failures
You know, I wonder if that high of a core voltage was messing with the on-die memory controller.
HT would be turned off or disabled for HFT platforms you need consistent fast computeing without possible different logic latency paths you would run two or more parallel code stacks and has race to win for for zero loss and the mtb would be either an Intel, supermicro, Dell. Fujitsu, oracle , HP, blades, could even been a Cisco device. IBM not some consumer retail boards. Get a server mtb, with no gpu just an on board vga
I bet it's some sort of firmware or memory IC issue. You would likely need the exact board meant to run this thing.
I had the same board and had the same problem with CD ram slots.
Agent47: der8auer you KNOW you weren't supposed to put this video out. * *Cocks silenced pistol* *
Seems like the CPU is toast. Electro migration from high voltage has taken another Processor. Thanks for trying with this limited edition proc Der8auer.
So this is the Black Flops.
Unlocked e5 2667v2 or e5 2680 v2 with disabled cores ? And no voltage limits she aint built for a longtime shes built for a good time kinda like a cuvic with an ebay turbo
That would have been hillarious if the black ops CPU worked on that "military class" board lol
Now this video gives me headaches of wanting some LGA2011 action...sure, I can get the CPUs from China or somewhere but finding a board will be another problem. Even if they'd run hot it will still be faster than the Core 2 Duo T7200 in a laptop I have...also running XP for older programs, newer boards doesn't have drivers for that
sounds to me like the memory controller on that cpu is unstable. try only 2 sticks or a single stick?
That tattoo keeps getting better 😅
that bsod reminds me of x58 dead imc symptoms, but since you tried the lowest clock single channel and everything it probably is a retail bios issue
I'd say if it could run at 1.5v for 10 years they would have just sold them like that. Guessing the reason they werent retail is running them that way isnt compatible with a retail warranty.
LOL! I KNOW THE FEELING!
Aren't hobbies supposed to be relaxing? This was more stressful than my actual job.