Friday, February 17th 2017
AMD Ryzen 1700X, 1600X & 1300 Benchmarks Leaked
A number of sites have been reporting on some leaked (as in, captured from Futuremark's database) scores on AMD's upcoming CPUs. Now, some benchmarks seem to have surfaced regarding not only the company's 8-core, 16-thread monsters, but also towards its sweet-spot 6-core, 12-thread CPUs and its more mundane 4-core offerings.
Taking into account some metrics (which you should, naturally, take with some grains of salt), and comparing Intel's and AMD's Ryzen offerings on 3DMark's Fire Strike Physics scores, we can see that a $389 Ryzen 7 1700X (8 cores, 16 threads) at its base clock of 3.4 GHz manages to surpass Intel's competing (in thread count alone, since it retails for $1089) 6900K running at its base 3.2 GHz frequency - with the Ryzen processor scoring 17,878 points versus the 6900K's 17,100. Doing some fast and hard maths, this would mean that if the R7 1700X was to be clocked at the same speed as the 6900K, it would still be faster, clock for clock (though not by much, admittedly). We don't know whether Turbo was disabled or not on these tests, for either AMD's or Intel's processor, so we have to consider that. However, if Turbo were enabled, that would mean that the R7 1700X's clockspeed would only be 100 MHz higher than the 6900K's (3.8 GHz max, vs 3.7 GHz max on the Intel CPU).
We see the same when comparing AMD's six-core, $259 R5 1600X against Intel's $617 6850K, with the Ryzen sample posting virtually the same score, despite running at a 300 MHz lower base clock (3.3 Ghz against Intel's 3.6 Ghz).
Jumping to a per-core analysis of processor speed in the same test suite, though, also reveals some very interesting metrics. here is a test which clearly doesn't scale all that well with extra cores, actually becoming more inefficient, per core, as the number of those increases. However, we can clearly see how much of an improvement AMD has achieved in per-core performance, with the R7 1700X scoring within spiting distance of its much more expensive i7 6900K competition.
Can we just get some real reviews of these pieces of silicon already?
Sources:
Videocardz, WCCFTech
Taking into account some metrics (which you should, naturally, take with some grains of salt), and comparing Intel's and AMD's Ryzen offerings on 3DMark's Fire Strike Physics scores, we can see that a $389 Ryzen 7 1700X (8 cores, 16 threads) at its base clock of 3.4 GHz manages to surpass Intel's competing (in thread count alone, since it retails for $1089) 6900K running at its base 3.2 GHz frequency - with the Ryzen processor scoring 17,878 points versus the 6900K's 17,100. Doing some fast and hard maths, this would mean that if the R7 1700X was to be clocked at the same speed as the 6900K, it would still be faster, clock for clock (though not by much, admittedly). We don't know whether Turbo was disabled or not on these tests, for either AMD's or Intel's processor, so we have to consider that. However, if Turbo were enabled, that would mean that the R7 1700X's clockspeed would only be 100 MHz higher than the 6900K's (3.8 GHz max, vs 3.7 GHz max on the Intel CPU).
Can we just get some real reviews of these pieces of silicon already?
99 Comments on AMD Ryzen 1700X, 1600X & 1300 Benchmarks Leaked
Performance difference between i7-6900K default frequency and OC is huge.
Not only CPU performance, memory write, read, copy, everything is improved a lot.
Overclocking Cache frequency is huge advantage of X99 platform and influence a lot on memory performance.
Now someone to ask me what you want i7-6900K or to wait AMD Ryzen with new board I would take i7-6900K without second thinking.You should sell i7-5960X, i7-6950X is enough. It's not wasted money if you can afford.
I would bought i7-6950X as well. Only I think his price is not real, Intel should stay in 1000$ range.
Before few days I saw one i7-6950X for 800 or 850 euro in my country. Excellent chance. But that's too much for me at the moment. I wait i7-6900K or i7-6850K for cheaper price.
I believe that AMD Ryzen could not beat score of overclocked i7-6900K. Both overclocked.
It may actually even match/beat it, we don't know yet.
Believe what you want, we should know the truth in a few weeks and making desicisions before all the facts are out seems a little backwards to me. If it was 6 months ago that would be another story.
if 1700X @ 4Ghz only getting 20300 CPU score, is great but not that good.
My 6900K @ 4GHhz getting 21900 CPU score.
Ram Mhz/CL is doing so much work at CPU Score, so we cant use this to a thing sadly.
What I want to know is how RAM speed will work with Ryzen. I got a sweet deal for 16GB of 3200 DDR4 memory and I'm betting on Ryzen to take some advantage of it. Have there been any reports about this?
If you are going to complain about the number of cores in Ryzen, you may as well complain about cars that have more than 200hp because most speed limits are under 65mph / 100kmh.
Second, about 9 months ago when I was looking, open drivers only supported Polaris and GCN 1.2. Not a biggie when buying a 480, but considering how AMD mixed architectures within a years' lineup, support was like a russian roulette.
Third, to this day the open driver is not on par feature wise with the closed driver. And the closed driver doesn't work with recent kernels.
So I had to pass.
Why the author didn't compare Ryzen to anything AMD currently offers?
There are some serious differences in the architecture and AMD always performance surprisingly well in some tests - even taking into account they've been using an outdated 28nm process.
Also, just from a realist standpoint, I deeply doubt the technological jump they would have to make to achieve such performance improvement, as this goes well beyond the obvious gain stemming from the 14nm process. Even when Intel and AMD used similar manufacturing tech, Intel was always extracting a bit more power from the silicon.
Nothing (e.g. takeovers or other know-how transitions) happened that could change this situation, but suddenly AMD shows a CPU that is rumored to be just as fast as Intel counterparts with similar power requirements and costing much less.
Also, we can expect Intel and AMD to be well informed on what the other company is developing at the moment. Just looking at the total laziness of Intel lately - crowned by the great joke called Kaby Lake - it seems they really aren't very worried about Ryzen. But while they most likely knew a lot about the performance and the manufacturing cost, they may have been surprised by the MSRP.
This instantly brings me to another idea: how many Ryzen CPUs does AMD expect to sell and how much are they willing to lose per chip, to regain some market share? Of course more CPUs mean more motherboards and possibly also more GPUs.
On topic Id like to see a real link to a real score, not someones self made graph.
PS
Intel, you're fired!
And since you don't know that Intel has been price hiking because they had no competition.....well that. AMD will not be losing any money at these prices. Intel has just been over charging for years. AMD seems to be realistic right now and are targeting markets they have little presence in like the enthusiast PC market and server market first. Once they have some traction there, they can focus on trying to take market share away from Intel. This should get their foot in the door so to speak.
They recently gave it to pentiums, making clock speed the only difference from 3mb cache i3's, and are just about to face an onslaught of amd chips with more than 4 threads in the i5's price bracket
From the first picture:
1700X @ 3.4ghz = 17,878 pts.
1700X @ 4.0ghz = 20,249 pts.
If we do your math (proportions), then the following is true:
17,878 pts (3.4 ghz)^-1 = n (4.0 ghz)^-1; n = a number with a unit in points for a score.
17,878 pts (3.4 ghz)^-1 (4.0 ghz) = 21,032.9 pts.
A difference in the score is 783.9 pts for 0.6 ghz gains.
There are a few points I would like to add.
1. From the first image, i7 6950x Broadwell-E seems to have a higher physic score, at stock, because it has more cores (10 cores, 20 threads). 1700X has 8 cores and 16 threads, and so it could be concluded that more cores improves this score. In addition, 1700X can exceed the score of the Broadwell-E if it is overclocked to 4.0 ghz, but if Broadwell-E was cranked up to 4.0ghz, its score would exceed by a large margin.
2. I don't understand the AMD-romance members are having about some Physics Scores related to Ryzen. I think it is meat for the masses and fanboys. Honestly, it doesn't tell the masses much, and Physic isn't necessarily needed for all PC games. The measurements only become relevant when members play PC games like Battlefield, The Old Republic, Call of Duty, Elder Scrolls Online, .... Basically the performance has meaning when physic calculations are being executed behind the rendering scenes of a FPS or game using it.
@mouacyk, I'm not arguing a point with you. Basically I am stating my point, but I am using your thread as start off point. So I hope you didn't take any offense of this.Broadwell-E doesn't OC that high in my opinion. I have my 6950x at 4.0 ghz. When rendering with VRay, it jumps into the low 70 deg C temperatures. If you compare it to Ivy Bridge-E 4960x, I had that processor up to 4.7, 4.8 ghz. It ran into the 90 deg C range when I use to render, and that setup was with a watercooling setup. Generally speaking, 4960x and 5960x will run faster than 6950x because of the difference in cores. The allure of Broadwell-E is you have a processor with 10 Cores. 10 Cores is ideal for a cheaper Xeon rig with no ECC memory, and you can render without spending $5,000 minimum on hardware alone. This is not including the cost for CGI software and 3rd party render nodes that are roughly $1,000 a license.
To answer your question, it will depend on the situation. If AMD came out with a 1900X with 10 cores, 20 threads at a higher clock speed, the answer would be yes and no. Yes because if you really care about performance, a theoretical 1900X could possibly beat a Broadwell-E. This thought is considering the scenario that Ryzen will live up to the hype. No you shouldn't worry because AMD doesn't always live up to the expectations, and AMD has a bad habit of setting the bar high for itself. It needs to set the bar high to compete with Intel, but AMD has a bad habit of falling a little short.
Ehm ehm, Pentium 4 had HT. It was huge thing back then, especially huge in not really working or being worth much. :D
HT isnt good for games, cause games lack ability to distinguish between real core and "just another thread". But its pretty good for stuff that scales with number of cores (eg. not games).
Also its reason why per-clock computation power of Ryzen is important. Gamers dont actually need multi-cores much. From what I play, it uses 2 cores in best case.. I could most likely live with high OC 2-core CPU in my games and never notice.
AMD in pre-Ryzen had a lots of multi-thread stuff but very poor actually computation power per-clock. Not much point in 5GHz CPU if its as effective as competition that has 3,2GHz.
Giving some CPU HT is about as useful for generic user as having old AMD. HT apart very specific stuff isnt actually good for ppl that dont know how to use it. Also due higher load per-core it tends to up watts consumed and heat created. Plus ofc due that it causes OC being less stable.
Off-topic of my reply..
AMD scales really strange way. I smell dead fish. :D
And HT can be good for games, just compare dual core chips with and without it, its more a case that games dont scale past 4 threads well, so i7 sees no benefit over i5
If these leaks are true and ryzen 8 core chips are about the same performance as intels most expensive overclocked or not, but for half the cost it's a win for AMD.
Low quality .gif for low quality leaks.