Question 14-core i9-10940X and the 12-core i9-10920X. Performance Leak on Geekbench.

nicalandia

Senior member
Jan 10, 2019
260
179
76
Two more 10th-gen Intel X-series processors leak on Geekbench, almost completing the lineup



Two more HEDT Intel CPUs rendered EOL by Zen2.....
 

Atari2600

Golden Member
Nov 22, 2016
1,201
1,298
106
****IF that is accurate****

Not a great outlook for Intel.

The 2990WX replacement won't have any of the NUMA and memory hopping issues that plagued the original.

Whatever about Zen2 not hitting the clocks that people wanted, fact is, it is still clocking nearly 10% higher than Zen+... which means if Intel has only stepped forward 10%, AMD have that mirrored in clocks - before you get to the 10+% ST IPC uplift and improved NUMA layout.

Then there is the rumour that TR3 *starts* at 16 cores. So the top of the line Intel HEDT i9-10980XE is taking heavy fire from the bottom of the line AMD HEDT. Not exactly good for perceptions is it?
 

Topweasel

Diamond Member
Oct 19, 2000
5,326
1,524
136
Untill you know the price you know nothing... What if it will be so cheap that it will have higher performace/price ratio than AMD offerings?
Okay lets look at this. 9900k is priced at $470, the 9900KS is still to come out is being priced at ~$600. Now sure this is supposed to be a limited edition super clocked chip "for gaming". But if we compare the Cascade Lake -X (do we have a short hand for this?). Just based on the 3900x you would need to price their 12c chip under $500. Do you see them lowering the intro price of their 10c Comet Lake down to $400, releasing a 10920x at $450 or lower, a 16c chip at under $700? When they are days away from a $600 8c chip?

I do believe that Intel would significantly drop prices. But once you realize how bad the pricing is, you get a better idea that there is so much cost grounds for Intel to catch up on, while being short on supply while making everything larger (or even doubling up) stuck on the same process. For example if they cut the 9920X's price in half its still $100 more expensive then the 3900x. It's almost like Server@home's jab with the EPYC review. With Intels pricing you almost have to discount them 80% to make them worth it. More like 65% here but its easy to see the point.
 

Kocicak

Senior member
Jan 17, 2019
279
149
76
Untill you know the price you know nothing...
Intel Cascade Lake-X
AnandTechCores
Threads
BaseAll
Core
TB2TB3TDPPrice
(1ku)
Core i9-10980XE18C / 36T3.03.84.64.8165 W$979
Core i9-10940X14C / 28T3.34.14.64.8165 W$784
Core i9-10920X12C / 24T3.54.34.64.8165 W$689
Core i9-10900X10C / 20T3.74.34.54.7165 W$590


See? It is actually cheap.
 

Atari2600

Golden Member
Nov 22, 2016
1,201
1,298
106
See? It is actually cheap.
Is it?

The bottom of the TR3 line will be a 16C CPU, which I would expect will trade blows with the $980 10980XE (due to better all core boost* and thermals**).

Pricing for previous bottom of the line Threadrippers:
1900X $550
2920X $650
39??X $???

I think its safe to say that the bottom of the TR3 line will come in considerably lower than $980.

*3.8GHz all core boost? My 2950X does better than that already.
**Does anyone really think these CPUs will come in at any thing like 165W? They will be pushing toward 300W!
 

Kocicak

Senior member
Jan 17, 2019
279
149
76
Is it?
I think its safe to say that the bottom of the TR3 line will come in considerably lower than $980.
**Does anyone really think these CPUs will come in at any thing like 165W? They will be pushing toward 300W!
Yes, for Intel, it is!
Until you know the price of TR3 you know nothing... ;)

I have been playing with Ryzen 1600 and 3600 recently and their 65W power rating corresponds to reality and really means something. Does Intel have anywhere in their white papers or specifications etc. burried the "real TDP" of their products? Or you just have to measure it yourself?
 

jpiniero

Diamond Member
Oct 1, 2010
7,924
1,206
126
Does Intel have anywhere in their white papers or specifications etc. burried the "real TDP" of their products? Or you just have to measure it yourself?
You could of course set PL2 to TDP if you wanted. Most boards by default now of course set it to 160 W/infinite time. It obviously won't boost as high.
 

Kocicak

Senior member
Jan 17, 2019
279
149
76
Oh. I believed that there is some "default setting" and if you plug a new CPU in MB, it will behave in a expected way. I do not have any other recent experience than with those two Ryzens and they behaved as I expected and both measured around 65 watts.
 

Atari2600

Golden Member
Nov 22, 2016
1,201
1,298
106
I have been playing with Ryzen 1600 and 3600 recently and their 65W power rating corresponds to reality and really means something. Does Intel have anywhere in their white papers or specifications etc. burried the "real TDP" of their products? Or you just have to measure it yourself?
I think you multiply the paper TDP by ( {[(max all core clock/ base clock)-1]*3} +1 ) or something to find Intels actual all cores flat out TDP.

For 9900K = (((4.7/3.6)-1)*3)+1 = 1.92; 1.92*95 = 182 W

For these:

real TDP
9900K185 W
990KS210 W
Core i9-10980XE300 W
Core i9-10940X285 W
Core i9-10920X280 W
Core i9-10900X245 W


*well, more postulate/guess. But its certainly more realistic than Intel's numbers!
 

Atari2600

Golden Member
Nov 22, 2016
1,201
1,298
106
Don't think there will be any less than the 24 core model, and I think that will be $999. Want less? Buy Ryzen.
I thought it was well established (Lisa Su?) that TR3 would start at 16 cores...?

i.e. https://www.tomshardware.co.uk/amd-threadripper-3000-castle-peak-16-core-cpu,news-61286.html

edit: For point of clarity, start != launch. They may launch with 24 (or more core) version; but I still expect the bottom rung of the TR3 ladder to be a 16C chip*.


*which is _______ amazing if you rewind to where we were in 2016.
 

TheGiant

Senior member
Jun 12, 2017
664
282
106
I think you multiply the paper TDP by ( {[(max all core clock/ base clock)-1]*3} +1 ) or something to find Intels actual all cores flat out TDP.

For 9900K = (((4.7/3.6)-1)*3)+1 = 1.92; 1.92*95 = 182 W

For these:

real TDP
9900K185 W
990KS210 W
Core i9-10980XE300 W
Core i9-10940X285 W
Core i9-10920X280 W
Core i9-10900X245 W


*well, more postulate/guess. But its certainly more realistic than Intel's numbers!
well thanks to the r1 and r2 lineup not able to achieve the clocks the difference between listed TDP and the reality was low, with r3 isnt
my r3900x while Handbraking (avx2) pulls 240-250W from the wall in default mode
as this is pure cpu workload, RAM, SDD and power losses (VRM, PSU) there is no way do 105W CPU to consume 105W while pulling 250W from the wall
TDP is a crap of nothing
to the topic
IMO cascade lake looks better now, but unless you have specific use case (4MemCH, PCI-E lanes, avx512) it is still not worth it
to the point in general, my 3900X is a no weak spot CPU (except the power I hate in 2019)
no gaming or multicore or single core weak spot
cascade lake X is security repaired skylake x, higher clock, more lanes, crap gaming
power I dont see an issue
the AMD fanboys will argue about 300W load, but that is while avx512 load and as I experience myself 3900X avx2 load there is nothing to add
I see skl-x lineup from 2017 as excellent CPU much better than the whole TR even considering the pricing, because of the clocks achievable
300W from the system of 7920X measures with 4ch and more lanes in 2017 at 4,6GHz vs 250W of 3900X means wasted 2 years of multicore time...
that pricing only reflects current situation
except when you are running cinbench (not4d) as your main workload./sarcasm off
 

Topweasel

Diamond Member
Oct 19, 2000
5,326
1,524
136
Intel Cascade Lake-X
AnandTechCores
Threads
BaseAll
Core
TB2TB3TDPPrice
(1ku)
Core i9-10980XE18C / 36T3.03.84.64.8165 W$979
Core i9-10940X14C / 28T3.34.14.64.8165 W$784
Core i9-10920X12C / 24T3.54.34.64.8165 W$689
Core i9-10900X10C / 20T3.74.34.54.7165 W$590
See? It is actually cheap.
Wow looking over this I really feel sorry for Intel. My 3900x is already pretty close to the 10920, the 4.6 is going to be really really compeitive considering the Cascade Lake -X's lowered IPC versus the 9900k and the general good IPC of Zen 2. I can see why they don't want to offer a 16c chip, the 3950x again not even HDET (well in AMD's HDET Portfolio, its pretty close in pricing and market), its going to run rough over that and put pressure on the 10980xe. Problem with Intel's design and why LGA 115x/1200 based chip do so well is the obvious smaller dies equals more binning, increased chance of cherry chips and so on. As anyone can see on the chart this HCC die really starts running out of steam as it specially as it gets to the 14c die. TR is only ever going to be thermally and power limited in its clocks as cores rise with the chiplets being more independent. I expect a 24c chip to clock similar to a 16c chip, and probably even the 32c chip as long as AMD is willing to offer them around 200w. Where Intel once had a clock advantage, the "Premiere" 24c chip is going to have 33% more cores and still probably clock at least as good as the 3900x.

Honestly if AMD doesn't offer a 16c TR at $700-$800 it will be a bit disappointing. They could make it a real blood bath. They might want to hold off on that one though to have it be a 4x0*4, so need stock of bad chips and to give people an opportunity to get impatient and get a 24c chip first (this has actually been their MO for a bit).
 

Atari2600

Golden Member
Nov 22, 2016
1,201
1,298
106
well thanks to the r1 and r2 lineup not able to achieve the clocks the difference between listed TDP and the reality was low, with r3 isnt
my r3900x while Handbraking (avx2) pulls 240-250W from the wall in default mode
as this is pure cpu workload, RAM, SDD and power losses (VRM, PSU) there is no way do 105W CPU to consume 105W while pulling 250W from the wall
TDP is a crap of nothing
I take it you measured system power draw at idle? and were not using the display heads at the time?

There were users elsewhere running a 3900X at 3.1 GHz fixed to prove that it got 65W - it did. Can't remember the workload - may not have been using AVX units.

It also doesn't tally up with results from reviewers.

y-Cruncher: 115W
Handbrake x264: 139W
Handbrake x265: 132W

Which is entirely in line with AMD's default PPT setting:
Package Power Tracking (PPT): The power threshold that is allowed to be delivered to the socket.
  • This is 88W for 65W TDP processors, and 142W for 105W TDP processors.
Do you have PBO enabled? [As at this point, stock TDP goes out the window, as can be observed within the review.]


Oh and I note you are inferring that TDP == Power consumed. It does not. It never has, but it is commonly misunderstood as being the same thing.
 
Last edited:

amrnuke

Senior member
Apr 24, 2019
841
1,025
96
How cheap would that have to be, though?
Well, it's workload dependent right? If we're audio processing, you could get a cheap XE and beat the 2990WX. If you're rendering, you want a TR for the most part.

Anyway, compiling results from Anandtech's review for HEDT performance (far from the be-all, end-all), the 2990WX loses to the 9980XE overall, and also relative to price, while the 2950X crushes them from a price-to-performance standpoint, with a 200%+ price to performance ratio. However, a 10980XE at $979... that's a totally different ball game. Price/performance it thus is expected to be on par with the 2950X, but with the overall performance still better than the 2990WX.

Really it depends on your use case, as above. But even at the previous price, the 9980XE wasn't exactly crappy from a sheer price-to-performance standpoint.

Long story short, this is good for us. TR3 had best be top-notch with the new pricing here.
 
  • Like
Reactions: Zucker2k

maddie

Diamond Member
Jul 18, 2010
3,159
1,829
136
Wow looking over this I really feel sorry for Intel. My 3900x is already pretty close to the 10920, the 4.6 is going to be really really compeitive considering the Cascade Lake -X's lowered IPC versus the 9900k and the general good IPC of Zen 2. I can see why they don't want to offer a 16c chip, the 3950x again not even HDET (well in AMD's HDET Portfolio, its pretty close in pricing and market), its going to run rough over that and put pressure on the 10980xe. Problem with Intel's design and why LGA 115x/1200 based chip do so well is the obvious smaller dies equals more binning, increased chance of cherry chips and so on. As anyone can see on the chart this HCC die really starts running out of steam as it specially as it gets to the 14c die. TR is only ever going to be thermally and power limited in its clocks as cores rise with the chiplets being more independent. I expect a 24c chip to clock similar to a 16c chip, and probably even the 32c chip as long as AMD is willing to offer them around 200w. Where Intel once had a clock advantage, the "Premiere" 24c chip is going to have 33% more cores and still probably clock at least as good as the 3900x.

Honestly if AMD doesn't offer a 16c TR at $700-$800 it will be a bit disappointing. They could make it a real blood bath. They might want to hold off on that one though to have it be a 4x0*4, so need stock of bad chips and to give people an opportunity to get impatient and get a 24c chip first (this has actually been their MO for a bit).
Why this, 'They might want to hold off on that one though to have it be a 4x0*4, so need stock of bad chips'?

These tiny die are probably yielding so high and of such low cost that AMD has no reason to leave any SKU off the table because of die harvesting delays. The margins on a 4+1 die TR 16C at $800 would be fantastic, even with the use of cut down more functional die.
 

Topweasel

Diamond Member
Oct 19, 2000
5,326
1,524
136
Well, it's workload dependent right? If we're audio processing, you could get a cheap XE and beat the 2990WX. If you're rendering, you want a TR for the most part.

Anyway, compiling results from Anandtech's review for HEDT performance (far from the be-all, end-all), the 2990WX loses to the 9980XE overall, and also relative to price, while the 2950X crushes them from a price-to-performance standpoint, with a 200%+ price to performance ratio. However, a 10980XE at $979... that's a totally different ball game. Price/performance it thus is expected to be on par with the 2950X, but with the overall performance still better than the 2990WX.

Really it depends on your use case, as above. But even at the previous price, the 9980XE wasn't exactly crappy from a sheer price-to-performance standpoint.

Long story short, this is good for us. TR3 had best be top-notch with the new pricing here.
Keep in mind that Zen 2 is the hammer here, can anyone name anything that a TR3 wouldn't clock for clock, core for core, and because of the design with the chiplets can clock almost as high if not higher as core count grows. The problem with Intel is they are only bringing the prices down to AMD's pricing for the last few gen, I don't see AMD cutting pricing to match with TR3. But at every core level AMD at HDET and higher it will be superior in just about everything. Gaming, check. Compute power, check. IO, check. If AMD even maintains pricing per core, Intel CPU's won't be terribly priced, but on every metric including $/perf TR3 wins.
 
  • Like
Reactions: lightmanek

amrnuke

Senior member
Apr 24, 2019
841
1,025
96
Keep in mind that Zen 2 is the hammer here, can anyone name anything that a TR3 wouldn't clock for clock, core for core, and because of the design with the chiplets can clock almost as high if not higher as core count grows. The problem with Intel is they are only bringing the prices down to AMD's pricing for the last few gen, I don't see AMD cutting pricing to match with TR3. But at every core level AMD at HDET and higher it will be superior in just about everything. Gaming, check. Compute power, check. IO, check. If AMD even maintains pricing per core, Intel CPU's won't be terribly priced, but on every metric including $/perf TR3 wins.
My point was that even at a $1979 price point for 9980XE and $1799 for 2990WX the 9980XE overall has better performance and better price-to-performance on those limited Anandtech HEDT tests. So this price drop for the 10980XE is huge. It makes the price-to-performance relative to the 9980XE (which was overall already better than the 2990WX) TWICE as good.

As for TR3 and 10920XE/10980XE, they haven't been released so frankly, no, no one can name anything that a TR3 would or wouldn't clock for clock, core for core, beat.

As for clock-for-clock and core-for-core, that's irrelevant. These aren't going into servers. All that matters for this is the performance for the task you're using it for, the power consumption, the total platform cost, and stability.

As an analogy, if someone gave me equal bets and payouts for Wilder vs Canelo in a boxing fight, I wouldn't say Canelo is going to win because pound-for-pound he's better. Of course he's going to get absolutely pummeled. I'd take Wilder 100% of the time.

If a theoretical 32C/64T 3990WX beats a 18C/36T 10980XE by 20%, at the same price and power consumption, I wouldn't claim the 10980XE is better... would you? So per-core performance doesn't matter.

Now, if a 10980XE at 3.0GHz -> 3.8/4.8 beats a theoretical 3990WX running at 3.2GHz -> 4.2GHz, are you going to call the 3990WX better because it has a lower boost? I hope not.
 
Last edited:

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,321
8,020
136
I agree. Pricing will be the deciding factor when TR3 releases. We are all pretty sure it will beat Rome core for core, due to higher clocks, and Rome just destroys the Xeon line, so TR3 will destroy this line. BUT AT WHAT PRICE !!!
 

Atari2600

Golden Member
Nov 22, 2016
1,201
1,298
106
My point was that even at a $1979 price point for 9980XE and $1799 for 2990WX the 9980XE overall has better performance and better price-to-performance on those limited Anandtech HEDT tests.
Stall the ball there.

The 2990WX is a very unique product, with NUMA levels all over the place.


When the memory hierarchy isn't a problem, the 2990WX absolutely flies. When it is a problem, you get performance significantly worse than the 2950X.

That'll all disappear with the Zen2 based TR3. Consistent performance due to the new memory subsystem topology.

I think you are making a bit of a mistake making such a general comparison between the two and then expecting that to extrapolate forward. Better to compare to 2950X and leave the two WX products aside - unless you are going to pick out benchmarks that show scaling in a linear fashion from 2920X thru 2950X to 2970WX and 2990WX.
 

Topweasel

Diamond Member
Oct 19, 2000
5,326
1,524
136
My point was that even at a $1979 price point for 9980XE and $1799 for 2990WX the 9980XE overall has better performance and better price-to-performance on those limited Anandtech HEDT tests. So this price drop for the 10980XE is huge. It makes the price-to-performance relative to the 9980XE (which was overall already better than the 2990WX) TWICE as good.

As for TR3 and 10920XE/10980XE, they haven't been released so frankly, no, no one can name anything that a TR3 would or wouldn't clock for clock, core for core, beat.

As for clock-for-clock and core-for-core, that's irrelevant. These aren't going into servers. All that matters for this is the performance for the task you're using it for, the power consumption, the total platform cost, and stability.

As an analogy, if someone gave me equal bets and payouts for Wilder vs Canelo in a boxing fight, I wouldn't say Canelo is going to win because pound-for-pound he's better. Of course he's going to get absolutely pummeled. I'd take Wilder 100% of the time.

If a theoretical 32C/64T 3990WX beats a 18C/36T 10980XE by 20%, at the same price and power consumption, I wouldn't claim the 10980XE is better... would you? So per-core performance doesn't matter.

Now, if a 10980XE at 3.0GHz -> 3.8/4.8 beats a theoretical 3990WX running at 3.2GHz -> 4.2GHz, are you going to call the 3990WX better because it has a lower boost? I hope not.
Obviously we don't know what the 10980XE performs like either? Wait we do.

Here are the facts as far as we know them about both.

1. Cascade Lake - X is a very slight process improved SL-X (as apposed to 9980xe which is) much like Coffee Lake is a process improved Skylake. All still 14nm
2. SL-X has reduced IPC vs SL due to mesh and changes made to boost specific server use cases (resembles Zen 1 in design in some choices).
3. For Yields as core count goes up even max ST turbo goes down.
4. Core for Core outside Spectre/Meltdown and a few other mitigations that Cascade Lake -x can be hardened from, SL-X and Cascade lake will be the same clock for clock. Which means we can project performance.

Here is what we know about Zen 2 through Ryzen and Epyc.

1. Memory performance and communication does not matter what CPU is talking to what channel.
2. That memory performance is drastically better then Zen and Zen 1 setups on EPYC.
3. That chip to chip communication is rather quick and actually inline with the mesh speed of SL-X.
4. That IPC of Zen exceeds that of Coffee Lake.
5. That while max clocks of 8c Ryzen does not quite match intels. The 3900x shows that they can as long as they are ok with the power and cooling can more easily maintain clock speeds as core count goes up.
6. That Ryzen 3k and EPYC shows that performance scales as linearly as possible as core usage goes up across any tool that can use it. Basically the IO for this example has no affect on performance. Whether its a 1 die, 2 die, or 8 die CPU. It performs as though it was one large Mono Die.

So what does this tell us about TR3.
1. That core speed no matter the core count up until general power limits, will keep up with Desktop Ryzen.
2. That non-uniform performances like the 2990wx are dead. A 32c TR3 will perform like a 16c TR3 and the 16c TR3 to a Ryzen 3k clock for clock (assuming mem bandwidth is not needed)
3. That it will hold a significant advantage in IPC over Cascade Lake because we see that cascade lake is worse then Coffee Lake and Zen 2 is better then Coffee Lake.

All we are missing is the exact clock info of TR3. Outside that we already know what it is. What it is, is going to be significantly faster then Cascade Lake - X. As long as AMD doesn't increase costs significantly per core over the TR2 pricing. Its going to offer far great $/perf. Then you add on the fact that Cascade Lake has to tap out before TR3 can ever start warming up the core count (I mean seriously. TR3 basically is at the lowest core count it can be at 16c, with Cascade Lake -X stopping at 18c).
 

amrnuke

Senior member
Apr 24, 2019
841
1,025
96
Stall the ball there.

The 2990WX is a very unique product, with NUMA levels all over the place.


When the memory hierarchy isn't a problem, the 2990WX absolutely flies. When it is a problem, you get performance significantly worse than the 2950X.

That'll all disappear with the Zen2 based TR3. Consistent performance due to the new memory subsystem topology.

I think you are making a bit of a mistake making such a general comparison between the two and then expecting that to extrapolate forward. Better to compare to 2950X and leave the two WX products aside - unless you are going to pick out benchmarks that show scaling in a linear fashion from 2920X thru 2950X to 2970WX and 2990WX.
Fair points. The 2950X is a better price to performance chip than either 2990WX or 9908XE. The 9980XE is no slouch, and isn't a particularly horrible value either, of course depending on workload. Everything about TR3 and 10xxxXE performance is all speculation. NUMA issues are important, and I look forward to seeing the performance of TR3. But... they're still issues that hamper the 2990WX.
 

ASK THE COMMUNITY