Tuesday, February 14th 2017

AMD Ryzen Benchmarks Leaked - Amazing Multi-core and Single-core Performance

Benchmarks have leaked on AMD's upcoming Ryzen CPUs, and if accurate, these are the ones that will change the name of the game from "Hype Train" to "Reality Check". Part of a verified Passmark entry, the test system consisted of an AMD Ryzen 8-core, 16-thread ES clocked at 3.4 GHz (which puts it closely on the Ryzen 7 1700X territory, though it isn't known whether Turbo to its rated 3.8 GHz was active or not), seated on an entry-level MSI A320 AM4 motherboard (absent of overclocking functionality) and 16GB of 2400MHz DDR4 memory.

The tests include integer math, floating point performance, prime numbers, encryption, compression, sorting, SSE performance and physics. The AMD Ryzen 7 1700X outperformed every other CPU in 5 out of the 8 tests, including Intel's fastest 8-core chip, the $1099 Broadwell-E i7 6900K. When put side by side against Intel's slightly less expensive $999 8 core extreme edition Haswell-E i7 5960X, Ryzen was faster in 6 out of the 8 tests. The 1700X showed particularly good performance in integer math and encryption, workloads typically associated with server workloads (and where the bulk of the profit is).
The average aggregate core Coming in at roughly 4% behind the i7-5960X and 9% behind the i7-6900K - both of which retail for more than double the pre-order price for the 1700X of $389. Consider that this isn't even the highest-clocked Ryzen CPUs about to hit retail, and things look rather good for AMD.
Moving on from the multi-threaded benchmarks, Passmark's single-threaded performance test is probably the most interesting one, due to AMD's recent inability to go toe to toe against Intel in single-core performance. This here was definitely the uphill battle for the company, and it would actually seem that Jim Keller and company have managed to do what might seem impossible, simply by looking at AMD's R&D budget deficit compared to its Goliath of an adversary in Intel. Whether or not the sample is running at stock 3.4 GHz sans Turbo (mightily impressive), or at the retail 1700 X's 3.8 GHz Turbo speed (still very impressive), the sample manages to successfully edge out the 5960X and the 6800K, falling behind the 6900K by no more than 3%. That's a very impressive feat, especially when one considers that Intel's i7-6900K and i7-6800K can Turbo up to 4.0GHz and 3.8GHz respectively in single-threaded mode, thanks to Broadwell-E's Turbo Boost 3.0 feature. This means that even if the Ryzen engineering sample was in fact running at 3.8GHz Turbo frequency, it would still be outperforming Broadwell-E clock for clock ever so slightly.
Only Intel's i7-7700K Kaby Lake with its 4.5 GHz Turbo manages to distance itself from the 1700X - a 91W 4-core going up against an 8-core chip, rated at only 4 watts higher at 95 W (rated; the jury is still out on real-world testing). And the power efficiency equation gets even more interesting (just not to Intel) when you consider that this Ryzen chip manages to come in at less 45 W TDP than Intel's 140 W i7-6900K. It would appear that AMD really did strike gold with the balance of features and power consumption, as well as multi and single-core performance.

If these benchmarks are real, the hype train has actually just vanished in smoke. Now it's called the Ryzen line, and it most likely represents a much-awaited supply train for AMD's pockets.Source: WCCFTech
Add your own comment

79 Comments on AMD Ryzen Benchmarks Leaked - Amazing Multi-core and Single-core Performance

#1
atomicus
Impressive though this is, I want to see actual game benchmarks and real world performance figures... benchmarks are by and large meaningless in the grand scheme of things. They may point to real world performance, but they are no hard core evidence of it. I await eagerly, and with optimism... hopefully disappointment will not follow.
Posted on Reply
#2
Liviu Cojocaru
Raevenlord said:
If these benchmarks are real, the hype train has actually just vanished in smoke. Now it's called the Ryzen line, and it most likely represents a much-awaited supply train for AMD's pockets.
Well put, let's hope that everything falls into place for AMD and we can finally see some real competition for Intel
Posted on Reply
#3
TheLaughingMan
atomicus said:
Impressive though this is, I want to see actual game benchmarks and real world performance figures... benchmarks are by and large meaningless in the grand scheme of things. They may point to real world performance, but they are no hard core evidence of it. I await eagerly, and with optimism... hopefully disappointment will not follow.
Man, who put salt in your coffee this morning?

The link to the original source was posted in another thread 2 days ago. I think they are real and close to Intel at that price is a win. Though I do agree that I want to see some real world game testing in CPU intensive titles with turbo ON.
Posted on Reply
#4
Eknex
Hypehypehype!

What exactly is Physics in the benchmarks? or what uses does it have?
Posted on Reply
#6
Rictorhell
IF all of this is true, all AMD has to do at this point is keep up with demand and they are golden. I am due for a new computer sometime in the near future so an AMD Ryzen purchase is definitely a possibility.
Posted on Reply
#7
HimymCZe
If AMD pull off that BS they did with RX400 cards, where they NDA everyone, then leaked faked report before release and DELIVER NONE OF THAT SHIT, I would consider myself DONE FOREVER with them.
Posted on Reply
#8
kruk
HimymCZe said:
If AMD pull off that BS they did with RX400 cards, where they NDA everyone, then leaked faked report before release and DELIVER NONE OF THAT SHIT, I would consider myself DONE FOREVER with them.
Links please ...

And also, nobody forces you to preorder hardware / buy it on release day.
Posted on Reply
#9
captainskyhawk
HimymCZe said:
If AMD pull off that BS they did with RX400 cards, where they NDA everyone, then leaked faked report before release and DELIVER NONE OF THAT SHIT, I would consider myself DONE FOREVER with them.
No links: FAKE NEWS. Sad!
Posted on Reply
#10
NdMk2o1o
HimymCZe said:
If AMD pull off that BS they did with RX400 cards, where they NDA everyone, then leaked faked report before release and DELIVER NONE OF THAT SHIT, I would consider myself DONE FOREVER with them.
I'm sure they'll be devastated :laugh:
Posted on Reply
#11
chaosmassive
AMD Ryzen indeed have HyperTrainding capable CPU !
Posted on Reply
#13
Atnevon
So curious to the Vets: Where do the Adobe CC benchmarks tend to lean? I'm replacing a 950 and I see a mix on those charts. I know an 1700X is foreseeable but I'm curious if the extra 100 bucks for the 1800X might take me 2 more years further down the road. A good game is a good game in all, and my GTX 1080 drivers are modded to run Mercury, Cuda, and all the Adobe GPU renders; just wondering if the extra umph will spread around nicely.
Posted on Reply
#14
TheMailMan78
Big Member
Ill wait for @cadaveca to review before I'm impressed.
Posted on Reply
#15
Mescalamba
1) where is proof they are real?
2) its ES, do we have some proof about freq? do we have some proof that ES will match retail?

IMHO, when something seems to be too good to be real, then its probably exactly that. I would really wish that AMD could really do something as solid as this, but Im really really skeptical about it.

And with this huge disappointment.. ehm, nevermind. :)

As for Physics test, I guess its PhysX for CPU? Still that benchmark doesnt make much sense, unless its somehow specifically tuned to Intel.
Posted on Reply
#16
Mescalamba
Atnevon said:
So curious to the Vets: Where do the Adobe CC benchmarks tend to lean? I'm replacing a 950 and I see a mix on those charts. I know an 1700X is foreseeable but I'm curious if the extra 100 bucks for the 1800X might take me 2 more years further down the road. A good game is a good game in all, and my GTX 1080 drivers are modded to run Mercury, Cuda, and all the Adobe GPU renders; just wondering if the extra umph will spread around nicely.
Might be just me but for me CC worked so far best with my generic Intel and nVidia combo. I had Intel and AMD (ATi) combo before and either drivers sucked or something was wrong with GPU or CC simply isnt tailored for ATi. Bottom line, it didnt work well and it actually had tendency to create graphical bugs. Might be better with current generation, but my last AMD GPU just wasnt "it" for this thing (was cool for GPU mining tho :D).
Posted on Reply
#17
P4-630
The Way It's Meant to be Played
Doesn't look to bad, if it's all true...
Who knows, these CPU's might be throttling around 55-60 degrees......:p
Posted on Reply
#18
Chaitanya



Mescalamba said:
Might be just me but for me CC worked so far best with my generic Intel and nVidia combo. I had Intel and AMD (ATi) combo before and either drivers sucked or something was wrong with GPU or CC simply isnt tailored for ATi. Bottom line, it didnt work well and it actually had tendency to create graphical bugs. Might be better with current generation, but my last AMD GPU just wasnt "it" for this thing (was cool for GPU mining tho :D).
even on current gen AMD gpus, CC doesnt work correctly. Its better to disable GPU acceleration or use nVidia GPU when using GPU acceleration.
Posted on Reply
#19
coolernoob
ok, AMD - at this point I am hyped beyond this:
just give to me already
Posted on Reply
#20
Atnevon
Mescalamba said:
I had Intel and AMD (ATi) combo before and either drivers sucked or something was wrong with GPU or CC simply isnt tailored for ATi. Bottom line, it didnt work well and it actually had tendency to create graphical bugs.
But in My case its a Ryzen CPU and an Nvidia GPU. No plans on replacing this GTX 1080 (now Hydro) any time soon. :D
Posted on Reply
#21
RejZoR
Even if these numbers turn out to be slightly overhyped, it's still a good start. Bulldozer was lagging behind from the beginning even to older gen Phenom II X6's... This one will be around on par with Intel offerings across the range. That's all consumers want. There is those 5% who want total and absolute king of the hill and the rest of 95% just want good options for decent money. And that's where Ryzen will shine, even if it's tiny bit slower than Intel in the end...
Posted on Reply
#23
jaggerwild
I don't actually "SEE" any bench mark scores(Am I F@@king blind as a bat!), which means hype but no prise, why don't they show any actual bench marks, like 3D mark or CPU-Z bench mark or or or? A link, why isnt RYZEN on HWBOT? And who is the go to AMD Over clocker that will get the honor of taking the first RYZEN to HWBOT?
Posted on Reply
#24
Prima.Vera
This is worst than the leaks for the new iPhone each year....
Posted on Reply
#25
Evo85
Don't expect HWBOT till the chip is released and its tested on a performance board.

The benchmarks posted are focused on the CPU. 3DMark is a graphics focused benchmark.

The jist of what has been released is to show the potential of these chips and how much they have closed the gap between Intel and AMD.

Take with a grain of salt and a sip of coffee.
Posted on Reply
Add your own comment