Alleged 6-Core Ryzen 7000-Series Tested in Basemark's GPU Rendering Tests

[TheLostSwede]

An AMD engineering sample CPU with the model name 100-000000593-20_Y has appeared in a couple of graphics rendering tests, paired with an NVIDIA RTX A4000 GPU. The CPU appears to be a 6-core Ryzen 7000-series chip that was fitted to a Gigabyte X670E Aorus Master motherboard. Based on the leaked information that was dug up by @TUM_APISAK, it has a clock speed of 4.4 GHz, but little else is known about the CPU. Basemark might not be the most widely used test out there and both the tests that were run, were GPU rendering tests. However, courtesy of @harukaze5719, we have some graphs comparing the alleged Ryzen 7000-series CPU with a Ryzen 9 5950X which is using the same GPU.

The first test is an OpenGL test where the 6-core CPU beats the 16-core CPU by a not insignificant 9.5 percent overall, but by more than 11 percent when it comes to the minimum frame rate in the benchmark. This is a significant performance lead, although in the Vulcan test, the difference is somewhat smaller with a 7 percent lead for the 6-core CPU. It's unclear how well these tests scale with more CPU cores, so we wouldn't read too much into either of these benchmark results, but it seems like AMD's Zen 4 architecture will deliver on what AMD has promised based on these early tests. We've independently verified that the numbers are in the ballpark of the engineering samples that AMD's partners have today, which means that these numbers haven't been faked in any way. Keep in mind that AMD is still working on its AM5 platform and it's still early days. We understand that AMD has recently fixed a few platform bugs that would've been showstoppers if AMD had launched the AM5 platform with them still present.



View at TechPowerUp Main Site | Source

 

[Wolverine2349]

DO these benchmarks scale with more CPU cores well??

The fact that it is a GPU rendering benchmark and many of those do not use many CPU cores at all (Even less than actual regulars games games) and the uplift is 7-11% is underwhelming and disappointing. I was hoping Zen 4 would be the more than 8 core with Golden Cove or better IPC I was hoping for. Maybe not??? Maybe still??

AMD Ryzen 9 7950X - Cinebench R23 Benchmark projection

We give you a preview of AMD's new Zen 4 desktop processors and calculate the performance of the AMD Ryzen 9 7950X top model for the new AM5 socket. The performance data and other technical details come directly from the press kit that AMD released as part of its Financial Analyst Day on June 9...

www.cpu-monkey.com

Per this it is on par with 12900K Cinebench single thread. And if it really needs 5.5GHz single core turbo boost to get there, that is not good at all. Now if it gets there at 5GHz, its pretty good.

Does anyone know if CPUs boost to single core turbo frequency by default when running Cinebench single thread test?? Or not always. I would think a CPU could boost to its rated 1 core frequency when only 1 core was used and if 5.5GHz was indeed used for 7950X single thread one core bench just to have parity with 12900K Golden Cove, that is very disappointing for sure.

 

[kapone32]

DO these benchmarks scale with more CPU cores well??

The fact that it is a GPU rendering benchmark and many of those do not use many CPU cores at all (Even less than actual regulars games games) and the uplift is 7-11% is underwhelming and disappointing. I was hoping Zen 4 would be the more than 8 core with Golden Cove or better IPC I was hoping for. Maybe not??? Maybe still??

AMD Ryzen 9 7950X - Cinebench R23 Benchmark projection

We give you a preview of AMD's new Zen 4 desktop processors and calculate the performance of the AMD Ryzen 9 7950X top model for the new AM5 socket. The performance data and other technical details come directly from the press kit that AMD released as part of its Financial Analyst Day on June 9...

www.cpu-monkey.com

Per this it is on par with 12900K Cinebench single thread. And if it really needs 5.5GHz single core turbo boost to get there, that is not good at all. Now if it gets there at 5GHz, its pretty good.

Does anyone know if CPUs boost to single core turbo frequency by default when running Cinebench single thread test?? Or not always. I would think a CPU could boost to its rated 1 core frequency when only 1 core was used and if 5.5GHz was indeed used for 7950X single thread one core bench just to have parity with 12900K Golden Cove, that is very disappointing for sure.

The 5950X (5.2 GHZ) is the fastest AM4 CPU in terms of single core. The fact that this is the replacement for the 5600X means that this bodes well for the CPUs above the 6 core variant. The 12900k is not a 6 cores CPU.

 

[Wolverine2349]

The 5950X (5.2 GHZ) is the fastest AM4 CPU in terms of single core. The fact that this is the replacement for the 5600X means that this bodes well for the CPUs above the 6 core variant. The 12900k is not a 6 cores CPU.

Does the 5950X really do 5.2GHz single core. I have seen its max advertised single core boost speed as 4.9GHz. I ahve seen some benches so it gets to 5GHz but never higher and only 1 core.

I hope your right. Though was the 7600X actually only running at 4.4GHz or was it boosting. And if you want to do an all core static overclock, is it going to gimp single thread much perf worse than Ryzen 5000. Cause with Ryzen 5000, I can get CCD1 on 5900X 4.7GHz all the time which is right up there to single core boost of 4.8GHz not losing much single thread performance.

Though is Zen 4 core going to really need 5.5GHz just to have parity with 12900K minus 8 points in Cinebench (though so small margin of error) in single core benchmark. And I know 12900K is an 8 excellent core CPU + 8 so-so/mediocre cores for 16 total, but that CInebench score was for single core only.

And the tests done here for 6 core 7600X Zen 4, how many CPU cores do they use given they were GOU rendering benchmarks,

I have run GPU rendering benchmarks like Unigine SuperPosition and it only uses 1 or maybe 2 CPU cores at most on my 5900X. Games on other hand will touch at least 4 cores and sometimes 6 and even sometimes 8 as they have more going on than just needing CPU IPC to drive a GPU engine benchmark like above.

 

[kapone32]

Does the 5950X really do 5.2GHz single core. I have seen its max advertised single core boost speed as 4.9GHz. I ahve seen some benches so it gets to 5GHz but never higher and only 1 core.

I hope your right. Though was the 7600X actually only running at 4.4GHz or was it boosting. And if you want to do an all core static overclock, is it going to gimp single thread much perf worse than Ryzen 5000. Cause with Ryzen 5000, I can get CCD1 on 5900X 4.7GHz all the time which is right up there to single core boost of 4.8GHz not losing much single thread performance.

Though is Zen 4 core going to really need 5.5GHz just to have parity with 12900K minus 8 points in Cinebench (though so small margin of error) And I know 12900K is an 8 real core CPU, but that CInebench score was for single core only.

And the tests done here for 6 core 7600X Zen 4, how many CPU cores do they use given they were GOU rendering benchmarks,

I have run GPU rendering benchmarks like Unigine SuperPosition and it only uses 1 or maybe 2 CPU cores at most on my 5900X. Games on other hand will touch at least 4 cores and sometimes 6 and even sometimes 8 as they have more going on than just needing CPU IPC to drive a GPU engine benchmark like above.

Depends on the board and if you do the 1 click OC in AMD software. 4.4 is not far away from the 4.6 that 1st gen 5600X could do. What we do not know is IPC, V-cache, Transistor density and a few other things that could determine what is going on. As an example of what I mean my 6500XT does 2958 MHZ on an air cooler while my 6800XT does not go over 2400 MHZ but the 6800XT is miles faster than the 6500XT. All I know if Dr Lisa Su is right and we can get all core 5 GHZ on these parts 5.5 GHz will not really be that important but would be nice and totally doable if they can bump the clock speeds. This is not the AMD of 2014.

 

[Wolverine2349]

Depends on the board and if you do the 1 click OC in AMD software. 4.4 is not far away from the 4.6 that 1st gen 5600X could do. What we do not know is IPC, V-cache, Transistor density and a few other things that could determine what is going on. As an example of what I mean my 6500XT does 2958 MHZ on an air cooler while my 6800XT does not go over 2400 MHZ but the 6800XT is miles faster than the 6500XT. All I know if Dr Lisa Su is right and we can get all core 5 GHZ on these parts 5.5 GHz will not really be that important but would be nice and totally doable if they can bump the clock speeds. This is not the AMD of 2014.

Do you think 5GHz will get single core Cinebench score between 1900 and 2000. Golden Cove 12900K (I eventually sold as I hate hybrid arch and no more than 8 cores) at 5GHz and I have tested this myself had a Cinebench single core score of somewhere in 19XX. Do you think Zen 4 single core 5GHz will match or exceed the 19XX of 5GHz Golden Cove??

 

[kapone32]

Do you think 5GHz will get single core Cinebench score between 1900 and 2000. Golden Cove 12900K (I eventually sold as I hate hybrid arch and no more than 8 cores) at 5GHz and I have tested this myself had a Cinebench single core score of somewhere in 19XX. Do you think Zen 4 single core 5GHz will match or exceed the 19XX of 5GHz Golden Cove??

As of right now no one can say. If the next generation is 15-20% faster than the current chips you will feel it regardless of the Cinebench score. The 5600G has the same boost clock as the 5600 but the 5600 is much faster.

 

[BlaezaLite]

I want to see Cinebench results...

 

[Wolverine2349]

I want to see Cinebench results...

AMD Ryzen 9 7950X Benchmark, Test and specs

The AMD Ryzen 9 7950X has 16 cores with 32 threads and is based on the 6. gen of the AMD Ryzen 9 series. The processor uses a mainboard with the AM5 (

www.cpu-monkey.com

Its 8 points behind 12900K though within margin of error. It lists CPU speed next to each CPU, but that is obviously CPU base speed not what the CPU running at as you can see 12900K is 3.2GHz and that is P cores base speed and obviously would not get 1997 Cinebench score at only 3.2GHz.

Do you know if CPUs auto boost to single core Turbo when running Cinebench single core/thread benchmark. I think 12900K P core single boost is 5GHz or around that.

It lists Ryzen 7950X has 5.5GHz. If it had to boost to that just to have parity with Golden Cove cores 5GHz, that is quite disappointing as it needed an extra 500MHz. I had a 12900K before that manual all core OC did 19XX something (forget exact numbers XX was) in Cinebench. SO I assume 5/5.1GHz for Golden Cove is what would get that Cinebench score or close to it.

Can you shed some light of AMD processors in general and also Intel boost to single core rated speed when doing single core Cinebench run. I am unsure as I usually do all core manual overclock.

 

[BlaezaLite]

AMD Ryzen 9 7950X Benchmark, Test and specs

The AMD Ryzen 9 7950X has 16 cores with 32 threads and is based on the 6. gen of the AMD Ryzen 9 series. The processor uses a mainboard with the AM5 (

www.cpu-monkey.com

When it's not a sample, then I'll be interested. I really can't see that being the results.

 

[Wolverine2349]

When it's not a sample, then I'll be interested. I really can't see that being the results.

Are samples usually worse than finally release even in IPC??

I do remember AMD with Zen 3, clock speeds in samples and demos were much higher than what final release was though IPC turned out to be sandbagging and IPC gains were great.

We will never know until final product.

 

[BlaezaLite]

I hope so, lol. I'm not a fanboy of AMD but I've become accustomed to the underdog pulling hard.

 

[Wolverine2349]

I hope so, lol. I'm not a fanboy of AMD but I've become accustomed to the underdog pulling hard.

I really hope so as well. I am not a fanboy of either company, but competition is good. Plus more importantly I really want as good or better than Golden Cove IPC and around 5Ghz or hopefully higher all core clock and more than 8 P cores. AMD is the only choice to turn to and only hope as Intel is just creating 8 P cores and more and more e-waste cores to CPUs which is why they are ruled out for me unless that changes. Or maybe the new Intel HEDT if it ever comes back and released to see the light of day and a 10-12 core CPU on a ring at that as a mesh arch is probably going to be much worse for gaming like Sylake-X was. I cannot stand the Intel Big.Little arch and being stuck at 8 P cores and them making it same Big.LittleXXXXX with next Gen with X being increase in e-crap atom cores and more hybrid arch crap.

If Intel had more 10 P cores on a ring that could clock 4900MHz all core or higher, I would have gone with that CPU.

 

[TheoneandonlyMrK]

If this got out I think it's a lowball personally, not by much but still, probably capping to a frequency.
Definitely a time for patient waiting, that Tpu review will come.

 

[Why_Me]

12900K is 16 core processor. You compare 6 cores vs. 16 .... a bit.. you know....

Isn't the AMD Ryzen 9 7950X a 16 core cpu.

 

[Wolverine2349]

12900K is 16 core processor. You compare 6 cores vs. 16 .... a bit.. you know....

Its 16 total cores but talking about 16 of same core type that is not a hybrid arch, but rather normal arch as CPU Monkey puts it under Core Architecture. It lists normal for Ryzen 5900X and Core i5 12400 and hybrid (Big.Little) for 12900K, 12600K and etc...

Isn't the AMD Ryzen 9 7950X a 16 core cpu.

Yes 16 of the same Zen 4 cores not a hybrid.

The CPU Monkey Cinebench results I was referring to is single core score only so it compares 1 Golden Cove core to one Zen 4 core and no more.

Trying to get an idea of how strong Zen 4 cores are compared to Golden cove. Imagine if they are as strong or stronger and we can get 16 of them. That should be a heck of a CPU.

 

[Mussels]

10% faster in single threaded performance would be a welcome generational uplift, if that holds for the 7000 series as a whole

Does the 5950X really do 5.2GHz single core. I have seen its max advertised single core boost speed as 4.9GHz. I ahve seen some benches so it gets to 5GHz but never higher and only 1 core.

I hope your right. Though was the 7600X actually only running at 4.4GHz or was it boosting. And if you want to do an all core static overclock, is it going to gimp single thread much perf worse than Ryzen 5000. Cause with Ryzen 5000, I can get CCD1 on 5900X 4.7GHz all the time which is right up there to single core boost of 4.8GHz not losing much single thread performance.

Though is Zen 4 core going to really need 5.5GHz just to have parity with 12900K minus 8 points in Cinebench (though so small margin of error) in single core benchmark. And I know 12900K is an 8 excellent core CPU + 8 so-so/mediocre cores for 16 total, but that CInebench score was for single core only.

And the tests done here for 6 core 7600X Zen 4, how many CPU cores do they use given they were GOU rendering benchmarks,

I have run GPU rendering benchmarks like Unigine SuperPosition and it only uses 1 or maybe 2 CPU cores at most on my 5900X. Games on other hand will touch at least 4 cores and sometimes 6 and even sometimes 8 as they have more going on than just needing CPU IPC to drive a GPU engine benchmark like above.

my 5800x is advertised as 3.8GHz to 4.7GHz, yet in reality the stock boost is 4850MHz - and PBO lets you raise that to 5050MHz (~4.4GHz all core with AVX loads)

AMD's specs are a little underwhelming when it comes to boost clocks, i honestly dont remember why - Nvidia does the same thing on ampere, with GPU's going way past advertised speeds

 

[Wolverine2349]

so called e-cores or efficiency cores are Atom processors(Atom architecture processors), just now rebranded as efficiency. Anyone who did have any product in the past with Atom processors in it knows what a bunch of sh*t is Atom processor.
hibrid not hibrid, still 16 cores

Yes very true. I should have wored more correctly to say an 8 good core processor. Because the 12900K to me would be better off as an 8/16 CPU as the e-cores slow the ring down create scheduling issues in Windows and are more of a nuisance than a help in today's day and age.

But yes they are real cores even if much weaker and it is a 16 core CPU.

ANd the unreleased Core i9 13900K is a real 24 core CPU as all 24 cores are real even though 16 of them are mediocre to subpar while 8 of them are excellent.

The thing is though consumers need to be aware that more cores is not always better. SOme may see 24 core 13900K and just think wow I get 24 cores for the same price of 16 without realizing that 16 of those 24 cores are much much weaker than all 24 cores in a Threadripper 3960X let alone the Zen 3 and then 4 equivalent.

Its all marketing and almost like a core count war. Though 9-10 years ago AMD had 8 core Bulldozer/Pilediver chips. Now some said those were not real cores, but oh yes they actually were. It would have been better to say they were real cores but awful crappy cores. Unknowing consumers thought 8 us better than 4 and bought those space heaters that had 8 horrendous cores with worse IPC than even original Conroe in a Sandy/Ivy/Haswell world where they had 50% IPC of those at same clock and sometimes worse and occasionally only a little better.

When something is real, but very crappy, it almost gets contagious to say not real even though it is because it feels useless to them. But yes they are real just mediocre/subpar overall e-cores and decent in limited scenarios. Though Bulldozers were actually worse and had no good cores to help, but different era when AMD was an afterthought.

 

[Minus Infinity]

Just LOL at those that think in ST Zen 4 will only be 10% faster than Zen 3. The IPC uplift alone is ~10%m the clock speeds are up to 14% faster, and AMD said themselves they were sandbagging about performance when they said > 15%. Sources are indicating 25%+ ST performance gains and much more in MT, so much so they still they can beat RL in MT despite it getting up to 8 additional peasant cores.

Any way really can't understand the angst for tests on probably engineering samples at low clock speeds. Same as I ignore any RL supposed benchmarks.

 

[Wolverine2349]

10% faster in single threaded performance would be a welcome generational uplift, if that holds for the 7000 series as a whole

10% overall including IPC and clock speed or just IPC. I would hope we have some modest clock speed improvements and at least a 10% IPC uplift.

my 5800x is advertised as 3.8GHz to 4.7GHz, yet in reality the stock boost is 4850MHz - and PBO lets you raise that to 5050MHz (~4.4GHz all core with AVX loads)

AMD's specs are a little underwhelming when it comes to boost clocks, i honestly dont remember why - Nvidia does the same thing on ampere, with GPU's going way past advertised speeds

Is that true do you you think just this gen. I remember hearing with Zen 2, that boost clocks were often overstated and not achieved.

Just LOL at those that think in ST Zen 4 will only be 10% faster than Zen 3. The IPC uplift alone is ~10%m the clock speeds are up to 14% faster, and AMD said themselves they were sandbagging about performance when they said > 15%. Sources are indicating 25%+ ST performance gains and much more in MT, so much so they still they can beat RL in MT despite it getting up to 8 additional peasant cores.

Any way really can't understand the angst for tests on probably engineering samples at low clock speeds. Same as I ignore any RL supposed benchmarks.

Yes true. How much lower clocked are engineering samples in general and are any potential boost clocks disabled??

 

[AlwaysHope]

...

Yes true. How much lower clocked are engineering samples in general and are any potential boost clocks disabled??

I've seen ES i9-12900K clocked at 1.8GHz here on ebay

 

[mtb scotland]

I want to see Cinebench results...

Why? they don't mean anything other than willy waving.

 

[Mussels]

Just LOL at those that think in ST Zen 4 will only be 10% faster than Zen 3. The IPC uplift alone is ~10%m the clock speeds are up to 14% faster, and AMD said themselves they were sandbagging about performance when they said > 15%. Sources are indicating 25%+ ST performance gains and much more in MT, so much so they still they can beat RL in MT despite it getting up to 8 additional peasant cores.

Any way really can't understand the angst for tests on probably engineering samples at low clock speeds. Same as I ignore any RL supposed benchmarks.

Never ever fall for the hype - If you see 25% gains listed, you can bet it'll be one niche example.
If we see a consistent 10% gain (in each segment, mobile, R5/R7/R9) we've got an upgrade - and if we get more than that, we're onto a fantastic series.

The problem is these days they just add more cores to claim faster in multi threaded, while ST remains pretty stagnant.

New laptop chips 50% faster, with 100% more cores! (meanwhile, gaming results don't get any faster despite adding the extra cores, they just uplift from the 100Mhz higher boost each model gets...)

 

[Wolverine2349]

Well said, except
..while 8 of them are excellent...
it's the same fallen behind Ryzon processor.
Intel's scam works like never before. They just mixed up their own not working (Atom) cr*p with lost in the battle and Victorious!!!

Are you saying that even Intel's 8 P -cores are same fallen behind AMD even Zen 3 CPUs??

Do you think IPC claims of being 10-15% better on Intel P cores compared to Zen 3 are exaggerated or real?

 

[Wolverine2349]

i think 45 nm is not a size of an atom.question myself how can Atom processor can be efficient if it doesn't run win11,win10,win8 and even win7?How??

I agree atom cores are crap and not power efficient except they seem only die space efficient which is still not a good excuse to use them. Just make a bigger socket and PCB for CPU. Would be even easier to cool as wider area to dissipate heat. Though the P cores are great performance and they could put more easily on a bigger die space.