Got recommend this video by the almighty algorithm, but am not disappointed. You went straight to the point and tests without rambling. Upvoted and subscribed.
This was quite interesting. I have a Xeon W5 2455x with 4x16GB dimms. I kind of want to do some benchmarks now with 32gb single channel, 32gb dual channel, 64gb tripple channel and 64gb quad channel
Hey! Great video! One quick comment though. I found it a little hard to see the Cinebench scores (on mobile). I think it’d be really helpful to have some on screen text for them. Keep it up!
Thanks for the feedback. Rewatching it now yeah those numbers are hard to see. For reference, the 16GB 1 channel had scores of: GPU: 222, CPU multi core: 330, CPU single core: 55. The 32GB 4 channel had: GPU: 3432, CPU multi core: 388, CPU single core: 50
Not entirely worthwhile unless you limit the number of changes made. Either the same amount of RAM (16Gb) but spread over 2 channels (or even 4x4Gb quad channel), or double the RAM but keep it single channel. Changing both means you can't know exactly where any improvements come from, and whether more RAM or Dual/Quad channels is a better upgrade, especially in tests like Fortnight where the before/after is so different. Also, what the hell are Lenovo thinking by shipping a xeon workstation with only single channel. SMH.
what's up pete? I am just researching this Lenovo P520, since I have seen that it is a very good option due to its low price and its high range of improvement, I have been thinking about buying one to build a good PC for games and productivity, with respect to the ram I have seen that in the official information on the Lenovo website this device is capable of reaching a maximum of 2933MHz in ram speed, you could try to test with modules at that speed and see how you do in the testing.
I highly recommend this PC! I still use it as my main workstation/gaming pc today. Thanks for the tip on the Ram speed, not sure if I will wind up testing that but good to know
I unironically wish consumer-grade chips had quad-channel memory controllers i know theres probably some technical bs that someone here is gonna know about more than i do that says its not practical like mobos being more expensive to support it etc, but i like to min-max preformance. I spent upwards of a dozen hours trying to get my ddr5 ramkit to be stable at 6200mhz for a better FLCK which improved my 0.1% and 1% lows in several games tested by roughly 2-5%>~ and about 2%~ for average fps, along with stress testing that with my CO present for my 7 7800X3D chip, even if my rams bottleneck is not a bottleneck in any of the scenarios i was doing (like I have to be playing microsoft flight simnulator on ultra settings or just a game that can write contious random iops that is gigabytes worth of information whatever that may be) to really see a difference worth caring about. Either way, its probably complete and total overkill for most case-use scenarios and more than enough for even what I just described, but a man could dream.
I do have a question on Fortnite. The stuttering indicates to me that it is loading up player models and skins. Do you have the options in the launcher set to pre-download them? (Same place you can download the “save the world” content). I just gave this a try on my 2010 Dell i7-2600 w/20 GB RAM, SSD, and nvidia 1060 6GB and did not have the same stuttering issue. The only time I encountered it was on the pre-load island and once I did a few 360’s it cached everything and stopped stuttering. Since I am rocking an ancient CPU and am 3 gens (4? Does 1660 count?) behind with 2 GB less video memory, I think the stuttering is something else than switching to quad channel or maybe even going from 16 GB to 32 GB (unless my 20 GB vs your 16 GB somehow makes the difference). I am running 1440p with higher graphics settings than “low” and I didn’t bother with an ftp overlay but in-game counter was 50-60 for my quick test. My “new” computer is a Dell workstation from 2015 (Xeon v4) and 3070, but I haven’t set it up fully to run a similar test to see about the stuttering issue you exhibit.
I don’t play Fortnite regularly, so it was my first time loading it in a while with the 16GB test. You are probably right it was loading in models, and then in the 32gb test everything was already loaded so seems like a one time thing. In the world the 32gb still had much higher fps though
Yes, though do you know if your model requires a 6 pin power connector? If so, you can do it, but you need a SATA to 6 pin connector cable. If it doesn’t require the 6 pin, then it will only need to draw power from the motherboard and no additional cable needed.
What do you have the fan stepping on? I swapped case fans to bequiets and added a second intake along with the noctua cooler and now added a second fan to that and man it runs so cool now. No more jet engine sound with the stock cooler and stock case fans. I can keep the fan stepping on 1 now also and it's so nice.
I still have the stock fans. They sound like a jet engine at startup but are fine after that. I’ve thought about adding some RGB fans (everyone knows RGB = more FPS right? 😂) but haven’t gotten around to it yet
@@PetesParts16 The noctua and bequiet are amazing and never sound like jets even under a crazy load. Adding a second intake also on the bottom of the front of the case helped temps so much and the 2 fans on the noctua cpu coooler are incredible. Noctua NH-D9DX i4 3U cpu coooler the second fan though had to go on left side due to all my ram slots are filled. But it was an instant temp drop and not a little one. Noctua NF-A9 PWM this fan added to cpu cooler, be quiet! Pure Wings 2 92mm the other 2 fans that work so well. It was worth it for overall temps. I wish I could remove the front storage/ usb and power buttons and add a third intake fan.
Thank you for making this video. I'm really sorry, but your comparison is so flawed that you probably should take down this video. Your major errors are: you compare RAM in single-channel usage against RAM in quad-channel usage without comparing the same amount of RAM. Especially in Fortnite, the jump from 16GB to 32GB of usage already makes a difference, as does the channel usage. You can't compare them side by side.
Now do it again with the 32GB config having 2 sticks per channel. Having more than one stick per channel is supposed to be omg horrible for performance and you have an opportunity to see if that’s true.
Depending on the quality of the dimms and the memory controller in the CPU they might not be able to run at their rated x.m.p speed, but otherwise there shouldn't be any penalty
@@PetesParts16 I would love to see that test. Mainly 32gb dual channel vs 32gb quad channel and see if the extra bandwidth does anything or of the performance comes from the capacity.
I just wish we had more than 2 channels on mainstream desktop at this point. It feels like an unnecessary limitation when we have so many cores today that are so fast. Surely they could use more ram bandwidth, or at least benefit from it.
Depending on what you're doing it definitely matters but that would make already expensive boards even more insane. A 7950X can already get limited by bandwidth. If desktop core counts keep going up, DDR6 will need to come sooner than later. I'd love to see triple channel make a come back like we had on X58 but I really doubt that'll happen on desktop since it doesn't make sense for gaming.
If you are GPU limited then it's not going to matter if you're running on an I with 128Gb of DDR6 or a Celeron with 4Gb of DDR3 as ... you are limited by the GPU
Everyone knows that one stick of RAM performs terrible on any computer. I would like to see a dual vs quad channel RAM comparison using both 1r and 2r ram. I don't think anyone has done a rank comparison before. I just can't find information regarding the best rank of ram to use with a 4 channel set up. This video looks like it was using 1r x8. Lots about a dual and none about quad channel ram because it is not common with gaming. I surmised that 2rx8 would be the best option based on some videos I watched but still am not sure.
Can you help point me in the right direction for purchasing ram? I just won one of these on eBay and it doesn’t have any ram. I’ve wanted to be able to build a pc for iracing for so long and I think this was my best option on the super small budget I have
That Xeon only has 8MB of cache so memory is even more important. Especially if ECC is enabled since that hits latency even more. Even my 6-core Xeon from 2011 had 20MB. The fact that they sell xeon workstations with one stick of memory like this is such a scam. They get businesses to pay top dollar then kneecap performance in hopes of getting them to upgrade sooner than they need to.
There's no such thing as a server PC... its a workstation pc. Like theres no remote management, no redundant PSU's, no rack mounting, nothing servery going on there... Anyway, I want to run the same test and see what I get with my 7960X, I suspect the lack of cache and ram speed having a dramatic affect in 1 channel vs quad channel.
Thr algorithm randomly sent me to you. Love the straightforward emperical experiment without tons of filler. Good job!
Thank you and thanks for stopping by!
Got recommend this video by the almighty algorithm, but am not disappointed. You went straight to the point and tests without rambling. Upvoted and subscribed.
Thank you, much appreciated!
This was quite interesting. I have a Xeon W5 2455x with 4x16GB dimms. I kind of want to do some benchmarks now with 32gb single channel, 32gb dual channel, 64gb tripple channel and 64gb quad channel
That would be an interesting experiment! I’d be curious to see those results
dual channel vs quad channel would be more meaningful.
Hey! Great video! One quick comment though. I found it a little hard to see the Cinebench scores (on mobile). I think it’d be really helpful to have some on screen text for them. Keep it up!
Thanks for the feedback. Rewatching it now yeah those numbers are hard to see. For reference, the 16GB 1 channel had scores of: GPU: 222, CPU multi core: 330, CPU single core: 55. The 32GB 4 channel had: GPU: 3432, CPU multi core: 388, CPU single core: 50
Not entirely worthwhile unless you limit the number of changes made. Either the same amount of RAM (16Gb) but spread over 2 channels (or even 4x4Gb quad channel), or double the RAM but keep it single channel. Changing both means you can't know exactly where any improvements come from, and whether more RAM or Dual/Quad channels is a better upgrade, especially in tests like Fortnight where the before/after is so different.
Also, what the hell are Lenovo thinking by shipping a xeon workstation with only single channel. SMH.
Someone else suggested testing with 4 sticks in 2 lanes and see how that compares which I may just have to do
what's up pete? I am just researching this Lenovo P520, since I have seen that it is a very good option due to its low price and its high range of improvement, I have been thinking about buying one to build a good PC for games and productivity, with respect to the ram I have seen that in the official information on the Lenovo website this device is capable of reaching a maximum of 2933MHz in ram speed, you could try to test with modules at that speed and see how you do in the testing.
I highly recommend this PC! I still use it as my main workstation/gaming pc today. Thanks for the tip on the Ram speed, not sure if I will wind up testing that but good to know
I unironically wish consumer-grade chips had quad-channel memory controllers i know theres probably some technical bs that someone here is gonna know about more than i do that says its not practical like mobos being more expensive to support it etc, but i like to min-max preformance. I spent upwards of a dozen hours trying to get my ddr5 ramkit to be stable at 6200mhz for a better FLCK which improved my 0.1% and 1% lows in several games tested by roughly 2-5%>~ and about 2%~ for average fps, along with stress testing that with my CO present for my 7 7800X3D chip, even if my rams bottleneck is not a bottleneck in any of the scenarios i was doing (like I have to be playing microsoft flight simnulator on ultra settings or just a game that can write contious random iops that is gigabytes worth of information whatever that may be) to really see a difference worth caring about. Either way, its probably complete and total overkill for most case-use scenarios and more than enough for even what I just described, but a man could dream.
Sooo, testing the speed of the actual RAM never crossed your mind?...
Nope sure didnt
I do have a question on Fortnite. The stuttering indicates to me that it is loading up player models and skins. Do you have the options in the launcher set to pre-download them? (Same place you can download the “save the world” content).
I just gave this a try on my 2010 Dell i7-2600 w/20 GB RAM, SSD, and nvidia 1060 6GB and did not have the same stuttering issue. The only time I encountered it was on the pre-load island and once I did a few 360’s it cached everything and stopped stuttering.
Since I am rocking an ancient CPU and am 3 gens (4? Does 1660 count?) behind with 2 GB less video memory, I think the stuttering is something else than switching to quad channel or maybe even going from 16 GB to 32 GB (unless my 20 GB vs your 16 GB somehow makes the difference). I am running 1440p with higher graphics settings than “low” and I didn’t bother with an ftp overlay but in-game counter was 50-60 for my quick test.
My “new” computer is a Dell workstation from 2015 (Xeon v4) and 3070, but I haven’t set it up fully to run a similar test to see about the stuttering issue you exhibit.
I don’t play Fortnite regularly, so it was my first time loading it in a while with the 16GB test. You are probably right it was loading in models, and then in the 32gb test everything was already loaded so seems like a one time thing. In the world the 32gb still had much higher fps though
Xeons support quad channel RAM. Core i3, i5, i7 etc and Ryzen do not support quad channel RAM
another big one check to see if resizable bar is enabled using gpuz
Hi, i have a question. Is the gigabyte geforce gtx 1650 compatible with a dell optiplex?
Yes, though do you know if your model requires a 6 pin power connector? If so, you can do it, but you need a SATA to 6 pin connector cable. If it doesn’t require the 6 pin, then it will only need to draw power from the motherboard and no additional cable needed.
well, older HEDT by intel were also quad channel, the lga 2011 socket was quad channel. had a 5820k myself for gaming.
What do you have the fan stepping on? I swapped case fans to bequiets and added a second intake along with the noctua cooler and now added a second fan to that and man it runs so cool now. No more jet engine sound with the stock cooler and stock case fans. I can keep the fan stepping on 1 now also and it's so nice.
I still have the stock fans. They sound like a jet engine at startup but are fine after that. I’ve thought about adding some RGB fans (everyone knows RGB = more FPS right? 😂) but haven’t gotten around to it yet
@@PetesParts16 The noctua and bequiet are amazing and never sound like jets even under a crazy load. Adding a second intake also on the bottom of the front of the case helped temps so much and the 2 fans on the noctua cpu coooler are incredible. Noctua NH-D9DX i4 3U cpu coooler the second fan though had to go on left side due to all my ram slots are filled. But it was an instant temp drop and not a little one. Noctua NF-A9 PWM this fan added to cpu cooler, be quiet! Pure Wings 2 92mm the other 2 fans that work so well. It was worth it for overall temps. I wish I could remove the front storage/ usb and power buttons and add a third intake fan.
Thank you for making this video. I'm really sorry, but your comparison is so flawed that you probably should take down this video. Your major errors are: you compare RAM in single-channel usage against RAM in quad-channel usage without comparing the same amount of RAM. Especially in Fortnite, the jump from 16GB to 32GB of usage already makes a difference, as does the channel usage. You can't compare them side by side.
Thanks for the view
Now do it again with the 32GB config having 2 sticks per channel. Having more than one stick per channel is supposed to be omg horrible for performance and you have an opportunity to see if that’s true.
Really great idea! I will have to test that out for sure
Depending on the quality of the dimms and the memory controller in the CPU they might not be able to run at their rated x.m.p speed, but otherwise there shouldn't be any penalty
@@PetesParts16 I would love to see that test. Mainly 32gb dual channel vs 32gb quad channel and see if the extra bandwidth does anything or of the performance comes from the capacity.
I am interested in knowing if 8 sticks is noticeably slower than four on this machine..
I just wish we had more than 2 channels on mainstream desktop at this point. It feels like an unnecessary limitation when we have so many cores today that are so fast. Surely they could use more ram bandwidth, or at least benefit from it.
Depending on what you're doing it definitely matters but that would make already expensive boards even more insane. A 7950X can already get limited by bandwidth. If desktop core counts keep going up, DDR6 will need to come sooner than later. I'd love to see triple channel make a come back like we had on X58 but I really doubt that'll happen on desktop since it doesn't make sense for gaming.
These thinkstations are a pretty good deal so you can always go that route if you want 4 channels
If you are GPU limited then it's not going to matter if you're running on an I with 128Gb of DDR6 or a Celeron with 4Gb of DDR3 as ... you are limited by the GPU
Is it possible to do a case swap on
the p520?
See reply on my other video 😎
Everyone knows that one stick of RAM performs terrible on any computer. I would like to see a dual vs quad channel RAM comparison using both 1r and 2r ram. I don't think anyone has done a rank comparison before. I just can't find information regarding the best rank of ram to use with a 4 channel set up. This video looks like it was using 1r x8. Lots about a dual and none about quad channel ram because it is not common with gaming. I surmised that 2rx8 would be the best option based on some videos I watched but still am not sure.
Compare 2 Channels to 8 channels
Can you help point me in the right direction for purchasing ram? I just won one of these on eBay and it doesn’t have any ram.
I’ve wanted to be able to build a pc for iracing for so long and I think this was my best option on the super small budget I have
Sure! You will need ECC ram since it’s a workstation PC. I put a link in the description of the 32gb RAM I went with in the video.
@@PetesParts16do you think I could make it work with the 2 8gb sticks for a total of 16gb?
@@WdotDynamite yes you definitely could do that
Awesome! Do you know what dimm slots I should use for just the 2 sticks?
Which ram are you using struggling to find compatible ram in the uk at reasonable price
The 32GB I upgraded with was Nemix Ram DDR4 2666 MHz
I've tried two now that's supposed to be compatible and they don't work driving me nuts lol
I wonder 32g dual (16gx2)ddr4 2666 vs 64g (16gX4)ddr4 2133 which one win
In theory should be 64gb but only a true test could tell!
@@PetesParts16 thank for info
That Xeon only has 8MB of cache so memory is even more important. Especially if ECC is enabled since that hits latency even more. Even my 6-core Xeon from 2011 had 20MB. The fact that they sell xeon workstations with one stick of memory like this is such a scam. They get businesses to pay top dollar then kneecap performance in hopes of getting them to upgrade sooner than they need to.
There's no such thing as a server PC... its a workstation pc. Like theres no remote management, no redundant PSU's, no rack mounting, nothing servery going on there...
Anyway, I want to run the same test and see what I get with my 7960X, I suspect the lack of cache and ram speed having a dramatic affect in 1 channel vs quad channel.
It can actually hold up to 512GB
Dude should really read up on installing ram on this system...