AMD Ryzen 2nd Generation Threadripper 2920X and 2970WX Review

[tinytomlogan]

If the 2nd Generation of Ryzen Threadripper CPUs piqued your interest but you wanted more cores, or fewer, then the 2920X and 2970WX should be on your shortlist.


Threadripper 2920X and 2970WX Review

 

[Damien c]

Tom not sure if it's just me or not but the video seems like it has some weird graphical issues like, a sharpness filter is on max and not working correctly and then the background having a few issues as well?

It's like the background is wobbling it's pretty weird to be honest, doesn't look like your normal quality of videos.

Either way I know my next upgrade is going to be on my render machine and it will be a threadripper, but still not sure if it will be 1st gen or 2nd gen threadripper.

 

[Wraith]

Not seeing it on my end, quality is as normal as any other video.

I'm actually quite tempted to go down the Thread ripper route for my next build.

 

[AlienALX]

Not seeing it on my end, quality is as normal as any other video.

I'm actually quite tempted to go down the Thread ripper route for my next build.

The boards are a little cheaper now, which was what made it easy for me to stick with X99 last upgrade.

Still no reason for me to upgrade here though. Well, it would be nice, but the cost is just far too high to make any sorta sense. Thing is, I took a risk on a 14 core CPU but looking around a bit? it's kinda starting to pay off. I just checked the specs for FO76 and you need quad core min with QC+HT recc. So we are heading in the right direction.

 

[Damien c]

Not seeing it on my end, quality is as normal as any other video.

I'm actually quite tempted to go down the Thread ripper route for my next build.

Strange will look at it later and see if it's because I was downloading although that normally doesn't affect it.

 

[AffenKatzen]

Temperature testing

Hi Tom,

I just wanted to start out with saying that I generally enjoy the content here and I value the reviews you guys do, but I was wondering what the reasoning behind having a graph dedicated to comparing temperatures of various CPUs when the test setup isn't kept identical?

As far as I can tell then only the results for 2950X, 2990WX, 2920x and 2970WX are from using the Coolermaster ML360 RGB TR4 AIO Cooler, whereas the i7-6700k used the Corsair H110i GTX and all other CPUs used the Corsair H110i GT.

In your review of the 2950X and 2990WX you explain the decision for using the coolermaster cooler with: "we're using the Coolermaster MasterLiquid ML360R RGB TR4 for our overclocking tests as there is no chance of keeping these monsters under control when overclocked using an air cooler.", but all other CPUs you compare with were already tested with an AIO, just 280mm instead of 360mm so that seems like an odd justification for changing from your regular setup.

So to go on and conclude that "The days of toasty AMD CPUs are long behind us and the Threadrippers, even with their enormous core counts, still remain cool under pressure." seems odd when you're specifically testing the threadripper CPUs with a beefier cooler than all others you compare it to.

If you didn't feel that your regular cooler, the Corsair H110i GT, would do a good enough job or you had other reasons for testing with the Coolermaster ML360 RGB TR4 then that's fine, but my point is just that the results shouldn't be in the same graph together when they aren't comparable.

 

[WYP]

Hi Tom,

I just wanted to start out with saying that I generally enjoy the content here and I value the reviews you guys do, but I was wondering what the reasoning behind having a graph dedicated to comparing temperatures of various CPUs when the test setup isn't kept identical?

As far as I can tell then only the results for 2950X, 2990WX, 2920x and 2970WX are from using the Coolermaster ML360 RGB TR4 AIO Cooler, whereas the i7-6700k used the Corsair H110i GTX and all other CPUs used the Corsair H110i GT.

In your review of the 2950X and 2990WX you explain the decision for using the coolermaster cooler with: "we're using the Coolermaster MasterLiquid ML360R RGB TR4 for our overclocking tests as there is no chance of keeping these monsters under control when overclocked using an air cooler.", but all other CPUs you compare with were already tested with an AIO, just 280mm instead of 360mm so that seems like an odd justification for changing from your regular setup.

So to go on and conclude that "The days of toasty AMD CPUs are long behind us and the Threadrippers, even with their enormous core counts, still remain cool under pressure." seems odd when you're specifically testing the threadripper CPUs with a beefier cooler than all others you compare it to.

If you didn't feel that your regular cooler, the Corsair H110i GT, would do a good enough job or you had other reasons for testing with the Coolermaster ML360 RGB TR4 then that's fine, but my point is just that the results shouldn't be in the same graph together when they aren't comparable.

TBH fella, with Threadripper you really need to be using a proper TR4 optimised cooler. Yes, standard Asetek mount coolers can work, but only proper TR4 coolers cover the full IHS and offer peak cooling efficiency.

As far as I am aware, Corsair doesn't even make any bespoke TR4 coolers, so apples-to-apples comparisons are impossible.

 

[AffenKatzen]

TBH fella, with Threadripper you really need to be using a proper TR4 optimised cooler. Yes, standard Asetek mount coolers can work, but only proper TR4 coolers cover the full IHS and offer peak cooling efficiency.

As far as I am aware, Corsair doesn't even make any bespoke TR4 coolers, so apples-to-apples comparisons are impossible.

If you agree with me that an apples-to-apples comparison isn't possible, then you must agree with me that the results should be in separate graphs instead of clumped together in one then?

 

[AlienALX]

From what I recall there are only two AIOs designed specifically for TR4. The Cooler Master and an Enermax. They are both 360s.

 

[Bartacus]

If you agree with me that an apples-to-apples comparison isn't possible, then you must agree with me that the results should be in separate graphs instead of clumped together in one then?

Not necessarily. The 6700k you referenced in your example doesn't produce anywhere near the heat level that these TR CPUs do, obviously. It's not an apples to apples comparison from the start, since you're comparing a desktop CPU to a 'workstation' CPU (for lack of better phrasing). The only question you need to ask is "during testing, was there thermal throttling". Using the same cooler for every CPU would be silly, especially these days, during the "core explosion" era.

 

[AlienALX]

I think it is a good thing. Let's put it this way, if you had *any* of the CPUs listed (whether apples to apples or not) then it would be highly likely you would be using the same cooler Tom did, or something pretty much identical.

So to me? I find it valuable information.

Ed. Also, this is not a cooler review it's a CPU review.

 

[AffenKatzen]

Not necessarily. The 6700k you referenced in your example doesn't produce anywhere near the heat level that these TR CPUs do, obviously. It's not an apples to apples comparison from the start, since you're comparing a desktop CPU to a 'workstation' CPU (for lack of better phrasing). The only question you need to ask is "during testing, was there thermal throttling". Using the same cooler for every CPU would be silly, especially these days, during the "core explosion" era.

The way the graph is setup invites comparison between the various CPUs when they are just listed in an identical manner and it says in the bottom of the graph "Temperature in C (lower = better)".

If you show someone that graph as it's presented and ask which CPU runs hottest they will naturally assume the 8700k followed by the 7900x and then 6700k. That is my issue with this.

In my last paragraph i specifically say that it's fine if they for whatever reason want to test specific CPUs with different coolers, but then results shouldn't be grouped together in a way which indicates that the results are comparable.

I think it is a good thing. Let's put it this way, if you had *any* of the CPUs listed (whether apples to apples or not) then it would be highly likely you would be using the same cooler Tom did, or something pretty much identical.

So to me? I find it valuable information.

Ed. Also, this is not a cooler review it's a CPU review.

I'm not saying it isn't valuable information how hot the CPU runs or that Tom isn't justified in changing to a different cooler, but that information could easily just had been presented without including the results from all the other CPUs which have been tested with a different cooler.

If the current graph was changed to one which only included the results for 2920x and 2970WX (or including the 2950X and 2990WX as well) then you would keep the relevant information without making comparisons which arn't justified.

 

[robbiec]

Hmm, reckon I could handle on of these in my rig (cooling / power wise), problem is that I don't really have a need for the extra compute over the 2700x... Diddums
Have a project at work on the go for an Edge Compute machine but we've been advised to use EPYC over Thread ripper so I'll be skipping on the middle rung.

 

[Deadtroopers]

Didn't actually listen to what you were saying Tom, your sheer animation and enthusiasm told me all I needed. I don't think even coffee by the pint would stimulate you so much! I'll listen later with the video off, so I'm not distracted as much, LOL!