Thursday, October 1st 2020

AMD Ryzen 9 5900X CPU-Z Bench Score Leaks, 27% Higher 1T Performance Over 3700X

With AMD expected to announce its 5th Generation Ryzen "Vermeer" desktop processors next week, the rumor-mill is grinding the finest spices. This time, an alleged CPU-Z Bench score of a 12-core/24-thread Ryzen 9 5900X processor surfaced. CPU-Z by CPUID has a lightweight internal benchmark that evaluates the single-threaded and multi-threaded performance of the processor, and provides reference scores from a selection of processors for comparison. The alleged 5900X sample is shown belting out a multi-threaded (nT) score of 9481.8 points, and single-threaded (1T) score of 652.8 points.

When compared to the internal reference score by CPUID for the Ryzen 7 3700X 8-core/16-thread processor, which is shown with 511 points 1T and 5433 points nT, the alleged 5900X ends up with a staggering 27% higher 1T score, and a 74% higher nT score. While the nT score is largely attributable to the 50% higher core-count, the 1T score is interesting. We predict that besides possibly higher clock-speeds for the 5900X, the "Zen 3" microarchitecture does offer a certain amount of IPC gain over "Zen 2" to account for the 27%. AMD's IPC parity with Intel is likely to tilt in its favor with "Zen 3," until Intel can whip something up with its "Cypress Cove" CPU cores on the 14 nm "Rocket Lake-S" processor.
Sources: 9550pro (Twitter), VideoCardz
Add your own comment

120 Comments on AMD Ryzen 9 5900X CPU-Z Bench Score Leaks, 27% Higher 1T Performance Over 3700X

#52
Prima.Vera
NeverdieMaybe it's time to upgrade my i5 3570k :p Don't know if the performance jump is worth it. :laugh:
I'm still on the i7 3770K. And since I play on 3440x1440 I see no reason to upgrade yet, since the CPU gain on that resolution is mediocre the best.
Posted on Reply
#53
Metroid
This is a hard pass if is indeed 15% ipc gain or higher, Intel days we had a 10% ipc gain and you could see people spending at least a thousand dollar on the upgrade because Intel demanded you to upgrade the motherboard too, in this case there is no motherboard upgrade, so it's easy to upgrade. I will have to say good bye to my ryzen 5 3600 and for the record, my 3600 does 500 on single-thread and 4000 on multi-thread, the new 5900 cpu does 652 on single-thread and 9481 on multi-thread. So if we cut that benchmark in half to 6 cores 12 threads, then we have single-thread still 652 and multi-thread 4740, which in turn makes 20% better multi-thread performance.
Posted on Reply
#54
Divide Overflow
I love my 3900X, but if single thread performance increases that much, I might upgrade the the 5900X.
I'll wait for the TPU review before making any decision.
Posted on Reply
#55
BorisDG
8 core vs 12 core ... interesting comparison. Yeah it's single thread, but still.
Posted on Reply
#56
Tomgang
This is seriously a great achievement for amd. 3700X has max boost of 4.4 ghz. Rumors say 5900X will achieve 5 ghz. With 27 % improvements the 5 ghz claims seems fairly legit I think. So 15 % IPC gain + 600 mhz higher coreclock = 27 % better single thread performance. This is not to far out in my opinion.

Also another leak where 5800X beating I9 10900K in Ashes of the Singularity benchmark. Gives good vibes for zen 3.

videocardz.com/newz/amd-ryzen-7-5800x-8-core-zen3-cpu-spotted-on-aots-benchmark-site

Unless Intel really brings something good to the table with 11 gen CPU's. It seems amd zen 3 is the go to choise for the coming time.

Now let's just hope scalpers dosesn't do a nvidia ampere again to Zen 3. Cleaning inventory in seconds.
Posted on Reply
#57
Space Lynx
Astronaut
TomgangThis is seriously a great achievement for amd. 3700X has max boost of 4.4 ghz. Rumors say 5900X will achieve 5 ghz. With 27 % improvements the 5 ghz claims seems fairly legit I think. So 15 % IPC gain + 600 mhz higher coreclock = 27 % better single thread performance. This is not to far out in my opinion.

Also another leak where 5800X beating I9 10900K in Ashes of the Singularity benchmark. Gives good vibes for zen 3.

videocardz.com/newz/amd-ryzen-7-5800x-8-core-zen3-cpu-spotted-on-aots-benchmark-site

Unless Intel really brings something good to the table with 11 gen CPU's. It seems amd zen 3 is the go to choise for the coming time.
all you need is my sig baby ~ Lisa Su has conquered all
Posted on Reply
#58
Tomgang
lynx29all you need is my sig baby ~ Lisa Su has conquered all
Zen 2 whas surprisingly good and now Zen 3 looks like a Intel killer even in games. Intel need to take amd very seriously now, else they will lose a lot of sales the way they are going now.
Posted on Reply
#59
HD64G
Intel in panic mode again in a few days. Who knows what else their marketing team will think of...
Posted on Reply
#60
Franzen4Real
NeverdieMaybe it's time to upgrade my i5 3570k :p Don't know if the performance jump is worth it. :laugh:
TomgangNow let's just hope scalpers dosesn't do a nvidia ampere again to Zen 3. Cleaning inventory in seconds.
Count on it. Now, botting and gouging are a trendy new topic at the forefront of tech conversations, right at the beginning of the Holiday spending extravaganza. I couldn't get a 3900X at launch or even months after for anywhere near msrp against human buyers, now we have freakin' automated 24/7 website scrapers to contend with as well. I think that there will be more people involved now that they have seen the $$ that can be made, at a time in the world where lost jobs and uncertainty has everyone wanting extra money and security more than ever. The high-end of Ryzen, RDNA2, Ampere, Xbox, PS5, you name it...if they are high dollar, highly anticipated items, then they are all F'd well through Christmas.

5900X is my goal, but I am anticipating disappointment through the end of the year minimum.
Posted on Reply
#61
Chrispy_
DrediI don’t think AMD can affect the XMP profiles memory manufacturers decide to put on their modules. Because of that the only way XMP could ”just work” on AMD is if they commit to designing their memory controller to be compatible with the same timings as the intel memory controllers, which is a silly proposition. The way forward is (sadly) QVL lists for memory, until memory manufacturers stop making their XMP settings incompatible with AMD memory controllers.
I'm not expecting XMP to work with Intel timings on an AMD motherboard, but AMD write AGESA firmware, and that includes speculative memory training.

At the moment, that memory training gives up on XMP timings far too easily because it doesn't loosen them enough. Even very loose 3600 timings and 1800 FCLK are waaaay better than JEDEC 2133 defaults.

Say you have a 3600 kit with 18-18-18-40 XMP timings that won't run on a typical Zen2 CPU; regardless of the board, the AGESA firmware will use the XMP primary timings and take a rough stab at the secondary timings before giving up. The thing is, XMP primary timings are almost always compatible with Zen2, it's the auto-generated secondary/tertiary timings that fail to boot on AMD and those aren't even part of the XMP spec. You can likely get that 3600-18-18-18-40 kit to run at 3600-16-16-16-36 on Zen2, so the problem is not the XMP data stored on the SPD.

IMO, AMD need to stick to the XMP frequency and voltage, and then use the primary timings as a reference point to calculate some safe values to attempt on the memory training runs. The best thing they could do at this point is hire 1usmus since his DRAM calculator works really well and is only a megabyte even as a compiled windows application with a GUI. If he can make a bootable timings calculator based off a handful of input variables, AMD can integrate the same kind of thing into AGESA. It doesn't even matter if AMD assumes low-quality RAM modules and runs a super-loose set of timings. Take that 3600 kit I mentioned above; Even if it was run at 20-20-20-55 timings with tRC of ~70 and tRFC of ~600 it would still be so much better than giving up and running the FCLK at 1066 instead of 1800 just because the memory training failed to find bootable values.
Posted on Reply
#62
Tomgang
Franzen4Real

Count on it. Now, botting and gouging are a trendy new topic at the forefront of tech conversations, right at the beginning of the Holiday spending extravaganza. I couldn't get a 3900X at launch or even months after for anywhere near msrp against human buyers, now we have freakin' automated 24/7 website scrapers to contend with as well. I think that there will be more people involved now that they have seen the $$ that can be made, at a time in the world where lost jobs and uncertainty has everyone wanting extra money and security more than ever. The high-end of Ryzen, RDNA2, Ampere, Xbox, PS5, you name it...if they are high dollar, highly anticipated items, then they are all F'd well through Christmas.

5900X is my goal, but I am anticipating disappointment through the end of the year minimum.
Yeah that's what worries me. Scalpers ruining my own Christmas presents a new pc.

Also about job. I lost my old back in April do to lock down and found a new job in July and have been there for 3 months. This week we have gone from pretty busy to almost nothing to do. If this continues, I can kiss this job goodbye as well. So I can only say I know to well how uncertainty can make people more desperate for money. It's not funny to go with out a job. Specially in the long run, as you feel the economic worries coming with low or no income at all situation.

Let's hope the vaccine will come fast, so we can get this dam virus out of the way and we can get back to more normal times. Althrow the virus is here to stay, but with a vaccine it will hopefully not be a bigger problem than the common cold/flu.
Posted on Reply
#63
TheoneandonlyMrK
Chrispy_I'm not expecting XMP to work with Intel timings on an AMD motherboard, but AMD write AGESA firmware, and that includes speculative memory training.

At the moment, that memory training gives up on XMP timings far too easily because it doesn't loosen them enough. Even very loose 3600 timings and 1800 FCLK are waaaay better than JEDEC 2133 defaults.

Say you have a 3600 kit with 18-18-18-40 XMP timings that won't run on a typical Zen2 CPU; regardless of the board, the AGESA firmware will use the XMP primary timings and take a rough stab at the secondary timings before giving up. The thing is, XMP primary timings are almost always compatible with Zen2, it's the auto-generated secondary/tertiary timings that fail to boot on AMD and those aren't even part of the XMP spec. You can likely get that 3600-18-18-18-40 kit to run at 3600-16-16-16-36 on Zen2, so the problem is not the XMP data stored on the SPD.

IMO, AMD need to stick to the XMP frequency and voltage, and then use the primary timings as a reference point to calculate some safe values to attempt on the memory training runs. The best thing they could do at this point is hire 1usmus since his DRAM calculator works really well and is only a megabyte even as a compiled windows application with a GUI. If he can make a bootable timings calculator based off a handful of input variables, AMD can integrate the same kind of thing into AGESA. It doesn't even matter if AMD assumes low-quality RAM modules and runs a super-loose set of timings. Take that 3600 kit I mentioned above; Even if it was run at 20-20-20-55 timings with tRC of ~70 and tRFC of ~600 it would still be so much better than giving up and running the FCLK at 1066 instead of 1800 just because the memory training failed to find bootable values.
How many different platforms have you used Ryzen calc on, it is not that great imho 18+ AMD builds.
AMD need their own Xmp ,intel timings are made for intel memory controller's which have different characteristics than AMD's.
Posted on Reply
#64
ARF
Why didn't AMD use 8-core CCX from the very beginning?
15% from Ryzen 9 3900X to Ryzen 9 5900X is nothing interesting, it's even disappointing given that they even jumped over one whole generation - Ryzen 4000 series.
Posted on Reply
#65
Calmmo
ARFWhy didn't AMD use 8-core CCX from the very beginning?
15% from Ryzen 9 3900X to Ryzen 9 5900X is nothing interesting, it's even disappointing given that they even jumped over one whole generation - Ryzen 4000 series.
Zen2 was big news for being 13% over zen+, we're used to more like 1-5% from intel (look at 9k vs 10k) and you find the potential of 15+% to be lackluster? erm.. I guess maybe it would have been in the early 2000's :confused:
Posted on Reply
#66
ARF
DemonicRyzen666I'm with this as long it's upgradable to DDR5 and Pcie 5.0.
I will skip this. Was also thinking about getting the new Big Navi but not going to happen since I have no good enough for me CPU to pair it with.

Waiting for DDR5 and PCIe 5, as well, maybe 2022, but at least it will be a strong and future-proof platform.

Hot and noisy X570 with last AM4 CPU , just no.
Posted on Reply
#67
TheoneandonlyMrK
ARFWhy didn't AMD use 8-core CCX from the very beginning?
15% from Ryzen 9 3900X to Ryzen 9 5900X is nothing interesting, it's even disappointing given that they even jumped over one whole generation - Ryzen 4000 series.
Wtaf , they skipped some numbers = butthurt.

They can call it Susan 43 I couldn't care less and they don't owe you two times the performance because they skipped a number.

Buy what you want obviously but spouting shite about 15% when it's looking like 27% and fifteen is above The norm anyway , go intel bro they'll give you more cores , performance and better bedroom heating, no wait only one of those is true.
Posted on Reply
#68
ShurikN
ARF15% from Ryzen 9 3900X to Ryzen 9 5900X is nothing interesting, it's even disappointing given that they even jumped over one whole generation - Ryzen 4000 series.
Pure comedy gold
Posted on Reply
#69
ARF
theoneandonlymrkWtaf , they skipped some numbers = butthurt.

They can call it Susan 43 I couldn't care less and they don't owe you two times the performance because they skipped a number.

Buy what you want obviously but spouting shite about 15% when it's looking like 27% and fifteen is above The norm anyway , go intel bro they'll give you more cores , performance and better bedroom heating, no wait only one of those is true.
Intel has a 10-core. AMD has an 8-core and far over it a distant 12-core.
Whatever.

It's called 1st generation, 2 generation, not 1st generation and then jump over to 5th generation.
It's misleading the customers who will think a second generation in advance.

Intel at least doesn't skip generations.
1st generation.
2nd, 3rd, 4th, 5th, 6th, 7th, 8th, 9th, and now 10th.
Posted on Reply
#70
siki
ARFI will skip this. Was also thinking about getting the new Big Navi but not going to happen since I have no good enough for me CPU to pair it with.

Waiting for DDR5 and PCIe 5, as well, maybe 2022, but at least it will be a strong and future-proof platform.

Hot and noisy X570 with last AM4 CPU , just no.
You two are bunch of losers because i am am waiting for PCIe 6. Yeah you heard it right.
And i also hope this CPU is compatible with PCIe 6. Yep thats me. Dumb as a rock.
Posted on Reply
#71
Calmmo
ARFfar over it a distant 12-core
so your agument is they should make the 12 core a 10core.
Sound, very.

You're really just trolling at this point with this whole skipping thing
Posted on Reply
#72
TheoneandonlyMrK
ARFIntel has a 10-core. AMD has an 8-core and far over it a distant 12-core.
Whatever.

It's called 1st generation, 2 generation, not 1st generation and then jump over to 5th generation.
It's misleading the customers who will think a second generation in advance.

Intel at least doesn't skip generations.
1st generation.
2nd, 3rd, 4th, 5th, 6th, 7th, 8th, 9th, and now 10th.
No they add a + reroll the dice and sell it again with 5-10% performance increases.
We got every AMD generation technically, just not an SKU for every segment.

And we're up to(intel gen) 11 next, 10900 etc are out ,the very easy to recall 11090 is next ,like who the f knows what Intel CPU is what ,they have a million different sku's a generation with some right turds hid about the place, Imho.

But back to your point they skipped a number, so what, dya really not recall Anyone else doing that.
Posted on Reply
#73
Chrispy_
theoneandonlymrkHow many different platforms have you used Ryzen calc on, it is not that great imho 18+ AMD builds.
AMD need their own Xmp ,intel timings are made for intel memory controller's which have different characteristics than AMD's.
Builds? I dunno; Low 3-digit figures myself, I have a couple of minions to crank out repeats and I just deal with sorting out new spec/builds/imaging when stocks dry up and force us to switch vendor/model etc. As for unique platforms, that probably spans at least a dozen B350/450/550/570 boards and three X399 models. When you have to keep ~1000 machines up to date on a 3-4 year cycle you tend to get through large quantities of builds and variants a month, but it's the personal commissions that actually give me just as much variety of build because there's no benefit in me standardising.

I reckon about 25% of the combinations, RAM doesn't work with the XMP profile, and I can always get it to work with the DRAM calc, though not necessarily with the default settings - sometimes I have to pick bad bin, or drop from B-die to OEM but it doesn't matter, I'm only concerned with getting them booting and stable at the higher FCLK and it's extremely rare (twice ever, I think) that I can't run at least the XMP's rated frequency and CL timing.

If you dig into what info the XMP 2.0 profile stores, very little of it is secondary and none of it is tertiary timings, yet these are the ones that need tweaking with bins or manual entry when the 1usmus safe defaults don't work using the presets, so it's not the XMP timings that are at fault, it's the poor attempts at memory training and ease with which motherboards/AGESA gives up and reverts back to JEDEC 2133.
Posted on Reply
#74
ARF
CalmmoZen2 was big news for being 13% over zen+, we're used to more like 1-5% from intel (look at 9k vs 10k) and you find the potential of 15+% to be lackluster? erm.. I guess maybe it would have been in the early 2000's :confused:
Intel increased the multi-threading performance from the 8-core 9900K to the 10-core 10900K by at least 25-30%.
Posted on Reply
#75
TheoneandonlyMrK
Chrispy_Builds? I dunno; Low 3-digit figures myself, I have a couple of minions to crank out repeats and I just deal with sorting out new spec/builds/imaging when stocks dry up and force us to switch vendor/model etc. As for unique platforms, that probably spans at least a dozen B350/450/550/570 boards and three X399 models. When you have to keep ~1000 machines up to date on a 3-4 year cycle you tend to get through large quantities of builds and variants a month, but it's the personal commissions that actually give me just as much variety of build because there's no benefit in me standardising.

I reckon about 25% of the combinations, RAM doesn't work with the XMP profile, and I can always get it to work with the DRAM calc, though not necessarily with the default settings - sometimes I have to pick bad bin, or drop from B-die to OEM but it doesn't matter, I'm only concerned with getting them booting and stable at the higher FCLK.
So Ryzen calc works the same there then, hit n miss, I could and did get by without it often too.
Patriot And certified Ram has been faultless for me though, even beyond Xmp ,last few builds I went with them, they're not expensive comparatively either.
But I still disagree that Xmp should just work, I understand it would be nice but it's unrealistic, didn't AMD make Amp an Xmp competition, not seen anything about it in years, shits vague now.

@ARF do the maths FFS, 8 cores add 2= 20% more cores = Intel having many a brew break chilling.

clap f#£@&g clap.
Posted on Reply
Add your own comment
Nov 29th, 2024 21:37 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts