AMD Ryzen Threadripper 3970X vs Intel Core i9-10980XE: High End Flagships Fight

gfg

Distinguished
Mar 30, 2005
93
27
18,670
I think it would be better to show the two processors alone and their overclock. In several games intel wins AMD clearly after overcloking and the conculusion does not accompany that.
 
Dec 19, 2019
2
6
15
I think it would be better to show OC results on a separate page altogether, much like its done over @guru3d. Including baseline and overclocks for the same chip in the same test just convolutes the graphs and makes it more cluttered than it needs to be. Most people arent going to buy a workstation CPU and OC to get a few more fps in games.
 

jdlech

Honorable
May 31, 2016
168
1
10,715
I'm starting to think there's a market for a chip that has decent multi thread performance, but has one specific processor with outstanding turbo boost... and a way to assign a demanding thread to that one processor.
It's OK that all the processors only go up to 4GHz. But what some programs really need is that one processor to crank it all the way up to 6GHz, even when all the surrounding processors slow down to compensate for the heat generation.
Because there's a lot of apps outside of FPS games that still rely on one single thread.
(looking at you, Stellaris).
 

Phaaze88

Titan
Ambassador
I think it would be better to show the two processors alone and their overclock. In several games intel wins AMD clearly after overcloking and the conculusion does not accompany that.
The majority of people of whom these processors are marketed towards aren't going to overclock them. They(overclockers) are the minority of PC owners.
Same deal for their use in a gaming PC. Only a handful of content creators would go that route, of which they'd all be well served with a 3900X/3950X/9900K + 2080Ti.

It's nice to see the OC potential of these chips, but it's not very important here.

I think it would be better to show OC results on a separate page altogether, much like its done over @guru3d. Including baseline and overclocks for the same chip in the same test just convolutes the graphs and makes it more cluttered than it needs to be. Most people arent going to buy a workstation CPU and OC to get a few more fps in games.
I agree that they should've been separated. It made the results a little confusing to look at, honestly.
Out of the box performance is key in this market.
 

joeblowsmynose

Distinguished
I think it would be better to show the two processors alone and their overclock. In several games intel wins AMD clearly after overcloking and the conculusion does not accompany that.

Considering that no CPU with more than 16 cores was ever designed as a gaming product, and that ANY person that needs 18 or 32 cores, isn't needing a 5% advantage in FPS, and isn't bottle-necking their CPU to game, perhaps the gaming results are irrelevant here.

Or perhaps there should be weighting of the values - for example, on HEDT CPUs, more weight to productivity scores at stock clocks, less weight to overclocking and less weight to gaming.

Where on say a 9700k / 3700x - less weight on productivity applications and more on gaming and overclocking.

That would be sensible.


But at the end of the day these TH "comparisons" are all a little tongue-in-cheek from my perspective -- don't take them so seriously.
 
  • Like
Reactions: TJ Hooker

ceomrman2

Distinguished
Apr 26, 2009
20
8
18,515
When comparing products head to head, it seems logical to roughly match either performance or price. The AMD is double the price of the Intel, plus its supporting cast is almost double, too. An entry-level workstation based on the Threadripper would cost almost twice as much as one based on the Intel. It would be much more capable, too. It's like comparing entry level cars with a Hyundai Accent and BMW 2-series. You'd have to drop all the way to a TR 2950X to get the CPU+MB price within 10%. That's a very different discussion. At this time, the Intel seems pretty logically priced in the marketplace. It's performance is clearly inferior to the latest TR chips, but that's why it's half the price.
 

dave.jeffers

Reputable
Nov 1, 2018
15
7
4,515
There really isn't a comparison between the Threadrippers and the Core i9-10980XE. This is why most reviewers chose to compare the i9 to the 3950X. Much closer performance, with the 3950X being much better value.
 
  • Like
Reactions: twotwotwo

Phaaze88

Titan
Ambassador
There really isn't a comparison between the Threadrippers and the Core i9-10980XE.
Do you realize you're denying the fact that both Threadripper and the 10980XE are marketed as HEDT parts?
Intel marketed the cpu as an HEDT part, therefore it gets evaluated as one. Threadripper is too, so there IS a comparison to be made between them.
The gaming results are just the bland side salad.

This is why most reviewers chose to compare the i9 to the 3950X. Much closer performance, with the 3950X being much better value.
3950X isn't a HEDT part. It's a mainstream one, albeit very expensive. It doesn't have the extra PCIE lanes or expandability of HEDT.

It is, and should be rather embarrassing for Intel when a mainstream part outperforms their HEDT one in professional workloads...
"You had one job... ONE JOB..."
 
  • Like
Reactions: joeblowsmynose
When comparing products head to head, it seems logical to roughly match either performance or price. The AMD is double the price of the Intel, plus its supporting cast is almost double, too.
Yeah, I never really like these "VS" articles here, as they tend to make questionable verdicts, the final numbers seem pretty arbitrary, and the matchups are sometimes less than ideal, but this seems like a particularly weird one. Just because these are the highest-end HEDT processors in each company's current lineup doesn't mean they should be compared against one another, as they are in totally different price ranges, and totally different performance ranges as far as multithreaded performance is concerned, which is what one would be buying these processors for.

The 3950X would probably be better comparison to put up against the 10980XE. Sure, it might not technically be on AMD's dedicated HEDT platform, but it's pretty much an HEDT part, surpassing AMD's prior Threadripper processors and competing with Intel's best current offerings in most ways. A 16-core, 32-thread processor at a $750 price point can hardly be considered a "mainstream" part at this point, and probably won't be for a number of years. And I wouldn't say a 3950X on an X570 board is far behind on the PCIe lane side of things either, considering the lanes provided by the CPU and chipset are all 4.0 lanes with double the bandwidth of their 3.0 counterparts. That's not exactly helpful when paired with 3.0 add-in cards, but may provide more benefit as time goes on.
 
D

Deleted member 2783327

Guest
Why do you people insist on comparing mainstream processors to HEDT processors. The 3950X and below have no place in this review. They don't have the PCIe lanes the HEDT CPUs do.
 

gaborbarla

Distinguished
I dont agree with the game test methodology here. The games seem to be cherry picked for high CPU core count and not actual average gaming perfomance.
Have a look here on steam for the most popular games: https://steamcharts.com/top

It would be lovely to include a few more mainstream games other than Aots and CIV VI, or hitman which I dont know anyone that plays. The article's gaming section should be titled: Games that are heavily CPU intensive performance. Try PUBG, DOTA2, CSGO, APEX legends, Fortnite. GTA V if you want to cover what a LOT of people play. This is my opinion, the list is flawed currently and does not reflect what people play and is skewed towards heavy CPU bottleneck than games that most people play.
 

Lkaos

Honorable
Dec 13, 2014
400
0
10,860
Is this for laughs? A CPU that costs almost double the other one and has a marginal fps (0.2 or .3) advantage is the better one!? Not in my book....
 

joeblowsmynose

Distinguished
Is this for laughs? A CPU that costs almost double the other one and has a marginal fps (0.2 or .3) advantage is the better one!? Not in my book....

I know funny right? They should have just eliminated the productivity tests altogether, and tested just games at 720p, because we all know everyone who buys a HEDT high core count CPUs only spent that money to play e-sport games. Why the hell else would you even buy a 32 core HEDT part? This comparison is so completely biased ...

Using productivity benchmarks for an HEDT comparison ... pffft!! What sort of crazy people even work at Tom's?



BTW re: gaming scores, considering that seems to be the only metric you understand -- with RAM tuned equally tight on 9900KS and Ryzen 3950x at 1080p with2080ti difference is only 4% - not far off from the difference (total avg) on this test ... and the one with that marginal fps advantage is the better gaming CPU? "Not in my book...."

If you want to use unreasonable logic, be prepared to have it applied both directions ... just sayin' ... :)

Besides ... some of you people area taking these "vs" tests way too seriously ... these articles aren't product reviews ...
 
Last edited:

joeblowsmynose

Distinguished
Do you realize you're denying the fact that both Threadripper and the 10980XE are marketed as HEDT parts?
Intel marketed the cpu as an HEDT part, therefore it gets evaluated as one. Threadripper is too, so there IS a comparison to be made between them.
The gaming results are just the bland side salad.


3950X isn't a HEDT part. It's a mainstream one, albeit very expensive. It doesn't have the extra PCIE lanes or expandability of HEDT.

It is, and should be rather embarrassing for Intel when a mainstream part outperforms their HEDT one in professional workloads...
"You had one job... ONE JOB..."


Just a note on the part I bolded ... it is embarassing - not just for Intel (strength of alignment dependent), and I would garner that is reason why the complaints ... if the 3950x was half as fast as the 10980xe ... I doubt anyone would complain. Notice no one complained about the 9900k in gaming benchmarks ... ?

"You had one job... ONE JOB..."

I almost spit out my coffee, thanks! :)
 
Last edited:
  • Like
Reactions: Phaaze88

Phaaze88

Titan
Ambassador
Is this for laughs? A CPU that costs almost double the other one and has a marginal fps (0.2 or .3) advantage is the better one!? Not in my book....
Comments like these are why the productivity and gaming benchmarks should be posted separately - or just drop the gaming benchmarks entirely, but NO, that's too tall of an order. WHY?
Because SO many people would be outraged at not seeing any gaming benchmarks... and most of those, if not all, don't even need cpus like these, but they're the main reason these benchmarks get put together anyways.

Like, I'm going to buy a Threadripper 3970X + 128GBs of ram + a T-Rex mobo + a Titan/Quadro/FirePro gpu(s) for games, for crying out loud... No, it's for work, so I can make some sweet cash, and if I have time on the side, I might play a round of Overwatch or LoL for example.
As if a few games aren't going to be unplayable, not that I'd care that much, because games wouldn't be my focus here.


Just a note on the part I bolded ... it is embarassing - not just for Intel, and I would garner that is reason why the complaints ... if the 3950x was half as fast as the 10980xe ... I doubt anyone would complain. Notice no one complained about the 9900k in gaming benchmarks ... ?
I noticed alright... there's some real biased folks out there...

I almost spit out my coffee :)
And I just finished mine for the morning, XD
 

joeblowsmynose

Distinguished
I noticed alright... there's some real biased folks out there...

I expect everyone to have some level of alignment one way or the other, but when one's alignment is so strong that it causes them to think and speak in unreasonable and illogical conclusions, people without such strong alignments look on and see only delusional people ...

People, don't be delusional ... the truth, logic, facts and objectivity are your best friends. Very strong alignments deplete those qualities. Writing stuff on the internet only exposes who you are (and yes, I am a sarcastic ass, thanks), so paint a pleasant picture of yourselves. :)
 

Makaveli

Splendid
Excellent review and thanks for taking the time to post it.

i'm going to agree with everyone that it maybe time to skip the games test or put them completely separate. The intel fan boys come out in droves when the blue team isn't on top. And the illogical reasoning is laughable at best on why one should be focusing on FPS in a product that is meant for work.

The amount of people buying this processors to game on will be almost close to 0.

The amount of people going to overclock these system will be even smaller than the group above.

Not difficult to understand unless you have some kinda bias.....
 
Dec 20, 2019
1
0
10
Alienware undoubtedly has some Mega Gaming $7K rig planned with one of these CPUs in it, never mind that it will be outperformed by something costing $3-$4k less...
I'm building a system that comes in around $6k (closer to 7 when I add cooling). How do you suggest I save $3-4k?
AMD Ryzen Threadripper 3970X 3.7GHz 32 Core sTRX4 Boxed Processor
EVGA FTW3 Ultra Gaming GeForce RTX 2080 Ti Triple-Fan 11GB GDDR6 PCIe Video Card
ASUS TRX40 ROG Zenith II Extreme AMD sTRX4 ATX Motherboard
Corsair Vengance RGB Pro 32GB 2 x 16GB DDR4-3200 PC4-25600 CL16 Dual Channel Desktop Memory Kit CMW32GX4M2Z3200 - Black
Samsung 970 Pro 1TB SSD 2-bit MLC NAND M.2 2280 PCIe NVMe 3.0 x4 Internal Solid State Drive
Corsair HXi Series HX1000 1000 Watt 80 Plus Platinum Modular ATX Power Supply
 
I'm building a system that comes in around $6k (closer to 7 when I add cooling). How do you suggest I save $3-4k?
Assuming you are talking about a system intended primarily for gaming, which is what they were referring to there, I wouldn't go with a 32-core, 64-thread HEDT processor where a majority of the cores are likely to never get utilized by games within the usable life of the system. I have some doubts that games will be making heavy use of more than 8 cores with 16 threads for some years, and one can get that with a $320 Ryzen 3700X on a ~$200 X570 AM4 motherboard. Or if you feel you must have more cores, there are 12 and 16-core models on the same AM4 platform for $500 and $750 respectively, but again, I wouldn't expect those extra cores to see much use in games for quite a while. And if it did turn out that certain games showed some benefit from those 12+ core processors down the line, you could always upgrade to one then, probably at a reduced cost.

Or go with a $500 i9-9900K or 9900KS with 8 cores and 16 threads with slightly better per-core performance in games, at least when paired with a high-end graphics card at relatively low resolutions. At high resolutions like 4K in modern games with the settings turned up, all of these CPUs should perform relatively similar though, as the 2080 Ti should be the limiting factor for performance most of the time.

I would not go with a $300 1TB SSD for gaming either, as it's not likely to improve game loading times noticeably over a model costing less than half as much. If I was to spend over $200 on an SSD, I would rather have one with double the capacity then one that gets slightly-higher performance, at least for a gaming build where a bunch of large games are likely to be installed.
 

nofanneeded

Respectable
Sep 29, 2019
1,541
251
2,090
I would not go with a $300 1TB SSD for gaming either, as it's not likely to improve game loading times noticeably over a model costing less than half as much. If I was to spend over $200 on an SSD, I would rather have one with double the capacity then one that gets slightly-higher performance, at least for a gaming build where a bunch of large games are likely to be installed.

If someone is willing to pay that much the good recommendation for him is to get Optane SSD , This will make HUGE difference in Load times.

Intel Optane 900p/905p
 

Makaveli

Splendid
If someone is willing to pay that much the good recommendation for him is to get Optane SSD , This will make HUGE difference in Load times.

Intel Optane 900p/905p

From what i've seen Optane doesn't improve gaming loading time much more so for applications so I wouldn't go that route if the only purpose is to play games.