Post your RYZEN bench scores! :)

[Morgan MLGman]

[done12many2]

5 minutes ago, Lays said:

wat was displaying the voltage?

 

cpu-z displays my volts wrong 99% of the time so maybe some trickery was involved

 Yeah, cpuz displays VID, but what was interesting is this was a proof of concept run for deciding a 7700k.  

 

It would run 5 GHz before and after the delid, but obviously the post delid runs were a lot cooler. 

[Lays]

1 minute ago, done12many2 said:

 Yeah, cpuz displays VID, but what was interesting is this was a proof of concept run for deciding a 7700k.  

 

It would run 5 GHz before and after the delid, but obviously the post reeled runs were a lot cooler. 

Are they available to like vendors and stuff yet? I know SL pretty well, if he has some I could probably ask for results of his findings

[done12many2]

13 minutes ago, Lays said:

Are they available to like vendors and stuff yet?

 

Not at all.

 

13 minutes ago, Lays said:

Are they available to like vendors and stuff yet? I know SL pretty well, if he has some I could probably ask for results of his findings

 

http://www.tweaktown.com/news/55497/7700k-delidded-30c-reduction-temps-wtf-intel/index.html

 

They were using Prime95 to generate temps.  I think 5GHz and higher might be pretty common with the 7700k.

 

I'll take a 15 min Prime95 run over a Cinebench run as proof of it running decent.

 

 

[done12many2]

24 minutes ago, rrubberr said:

These figures just aren't possible from a CPU. My laptop's GPU, a 1070, running at 1911 MHz, scores eight seconds. They must be benching their integrated graphics or something, as there's no way in HECK they have an x86 CPU pushing 3 TFLOPS.

 

What are you talking about?

[done12many2]

6 minutes ago, rrubberr said:

Eight seconds on a CPU isn't a realistic number.

 

If you are achieving an eight second run with the Blender file provided by AMD, you have adjusted the sample setting or something else.

 

Go here to download Blender version 2.78a for Windows.  Their comparison was done on Windows 64-bit so anything other than that is not directly comparable.

 

Go here to download the file from AMD to render.   Follow the instruction, but to some it up, don't touch any and make sure sample is set to 150.  Run test.

 

If you do it correctly there is no way you'll come even remotely close to an 8 second render.  Probably like a minute plus or so.

 

[done12many2]

3 minutes ago, rrubberr said:

http://www.pcworld.com/article/3151464/hardware/you-can-find-out-how-your-cpu-compares-to-amds-ryzen-for-free.html

 

The point of this is comparing a GPU, which is notoriously GOOD at floating point, to a CPU, which is NOT without integrated graphics of some sort. I already had the latest version of blender, I use it almost daily for work. I downloaded and ran the benchmark without changing anything save for the render device.

 

In regards to your last line, that's the point I was trying to make.

 

The point is NOT to compare GPUs, but CPUs.  I highlighted that in the link you provided as well.

 

It doesn't mention GPU one time in the link you provided either.

[done12many2]

Just now, rrubberr said:

The POINT I'm trying to make is that AMD did not SHOW A CPU TIME.

 

It's 35 seconds for the Ryzen CPU to render the provided job at 150 samples.  How did you even get on the topic of GPU render times?

[done12many2]

Just now, rrubberr said:

Oh, sorry, nvm, I though thy were showcasing a consumer grade quad core octa thread model. In that case, my argument flips 180 degrees. Sorry.

 

No worries man.  Give it a spin when if you have some time.  It's always interesting to see how our own rigs compare to new technologies.

[asand1]

[Clanscorpia]

On 12/16/2016 at 7:27 PM, Jorgen297 said:

Ryzen will PWN any intel CPU

no duh its the highest end CPU

[X_X]

Umm, maybe I should try to offload my i7-6800k while prices are still good.

[PianoPlayer88Key]

5 hours ago, SpaceGhostC2C said:

I don't know where you took those numbers form, but Piledriver has slightly better IPC than Phenom II, which was substantially better than Core 2 (so was Nehalem).

 

In any case, I wouldn't expect AMD to surpass intel in IPC, since not even AMD dares to claim that (they keep focusing on whether they match Haswell-E and Boradwell-E, which are behind Skylake in terms of IPC).

The good news is that we only need AMD to be in the same ballpark as intel, even if somewhat behind, to influence pricing (who wouldn't take a 5% slower CPU than top of the line if its price is sufficiently lower? i5s exist after all). A competitive, even if worse, quad-core by AMD would probably bring the mainstream i7s back to reason, which could imply the end if i5s as we know them (my back-of-the-envelope guesstimate is that i5 pricing is what i7 pricing would look like in a competitive market, since cost-wise the difference is minimal).

I was looking at the single-threaded scores from passmark and userbenchmark of CPUs from several generations.

For example, on userbenchmark, the 2.66 GHz i5-750 has 4% better single-threaded performance than the 3.8 GHz FX-4300, the 2.66 GHz i7-920 is the same as the 3.5 GHz FX-8320, the 3.4 GHz i7-2600K is 44% faster than the FX-8320, the i7-920 equals the 3.7 GHz X4 860K, and the 2600K is 45% faster than the 860K.

On passmark (cpubenchmark), going by single-core scores and turbo clocks, the FX x3xx CPUs appear to be slightly edged out by the Conroe Core 2 Duos and Kentsfield Core 2 Quads in single-threaded IPC.

 

I'd like to see the price wars we had like 10 or so years ago, IIRC.  I can't access the site right now (I get a 403 error) but I remember looking up the Q6600 on cpu-world, and it started at like $850 and after a while dropped to like $250 or something like that.  I hope we have similar or larger (percentage) price drops over on the blue side - for example the 6950X dropping to like ~$500, 6900K to ~$325, 6850K to ~$200, 6800K to ~$130, 6700K to ~$105, 6600K to ~$75, and all dual-core (including hyperthreaded chips) discontinued, including mobile.

 

That, and I'm tired of marginal per-generation performance improvements.  I really want to see them return to the 8086 -> 286, 486 -> Pentium (I've heard the preceeding 2 were like 2x IPC at the same clock speeds, all single-core remember), Core 2 -> Nehalem improvements, or beat the Kepler -> Maxwell -> Pascal generational improvements.  Better yet, if they'd catch up to where they would be if the fastest-ever rate of per-generation or per-year IPC improvements had not ever slowed, and preferably beat even that.

I like to see a 3-5x improvement in single-threaded IPC, or a 10-16x or more improvement in overall performance, at the same price and TDP before I change motherboards/sockets, and I don't want to have to wait 5, 7, or even 10 years (from buying my 4790K in January 2015) to get that.  (Also I'd like to see 2x single-thread, 4x overall improvement or better at the same price/TDP on the same socket in the future.)  I just don't like swapping out motherboards a lot.  If it was as easy/quick as replacing a video card or stick of RAM, I'd feel different.   (It's the labor involved, not the purchase price, that bothers me.)

[Masteryodi1980]

1 hour ago, X_X said:

Umm, maybe I should try to offload my i7-6800k while prices are still good.

Where was you nearly 3 months ago when I was in the market for one.......?!

[Kc7vwc]

i5 4690K using only MSI OC Genie = 1:35.

[SpaceGhostC2C]

51 minutes ago, PianoPlayer88Key said:

I was looking at the single-threaded scores from passmark and userbenchmark of CPUs from several generations.

For example, on userbenchmark, the 2.66 GHz i5-750 has 4% better single-threaded performance than the 3.8 GHz FX-4300, the 2.66 GHz i7-920 is the same as the 3.5 GHz FX-8320, the 3.4 GHz i7-2600K is 44% faster than the FX-8320, the i7-920 equals the 3.7 GHz X4 860K, and the 2600K is 45% faster than the 860K.

On passmark (cpubenchmark), going by single-core scores and turbo clocks, the FX x3xx CPUs appear to be slightly edged out by the Conroe Core 2 Duos and Kentsfield Core 2 Quads in single-threaded IPC.

But if you used userbenchmark, the q6600 vs Piledriver comparison would give you ~100% higher SC performance with less than double the frequency (67-75% higher)  

Even the Phenom II 945 scores 44% higher at 3.0, and the big complaint about Bulldozer was that it was equal or slightly worse in IPC than Phenom, with Piledriver being slightly better...

Anyway, I'm digressing, the topic wasn't precisely how old CPUs perform (at least not that old ).

 

1 hour ago, PianoPlayer88Key said:

I'd like to see the price wars we had like 10 or so years ago, IIRC.  I can't access the site right now (I get a 403 error) but I remember looking up the Q6600 on cpu-world, and it started at like $850 and after a while dropped to like $250 or something like that.  I hope we have similar or larger (percentage) price drops over on the blue side - for example the 6950X dropping to like ~$500, 6900K to ~$325, 6850K to ~$200, 6800K to ~$130, 6700K to ~$105, 6600K to ~$75, and all dual-core (including hyperthreaded chips) discontinued, including mobile.

That also depends on what launch prices were. I mean, whether they were set realistically from the beginning, or they were just some random high number to impress people, and then reality kicked in  For example, most AM3+ CPUs have experienced a mild decline, if any, for years, and not at EOL we some steady downward trend. The 9370 and 9590, on the other hand, were announced at absurd prices nobody cares about (I guess we can't rule out one or two persons with too much money and not enough clue), and then at some point they got a real price at which to be sold (in the 9590 case, much less than 50% the original price). I think (speculation detected) $200-250 is the price of a good quad core these days, i5 and i7 being just the artificial result of price discrimination by a monopolist. I would expect a successful Zen 4c to land at that level, at the higher range while Intel maintains its pricing, then eventually both Zen and i7 coming to $200 if competition is real.

 

Quote

That, and I'm tired of marginal per-generation performance improvements.  I really want to see them return to the 8086 -> 286, 486 -> Pentium (I've heard the preceeding 2 were like 2x IPC at the same clock speeds, all single-core remember), Core 2 -> Nehalem improvements, or beat the Kepler -> Maxwell -> Pascal generational improvements.  Better yet, if they'd catch up to where they would be if the fastest-ever rate of per-generation or per-year IPC improvements had not ever slowed, and preferably beat even that.  

That's a different story. AMD was even more tired of sucking at IPC, but a new, substantially better architecture is not something you can come up with every day. It's more than a matter of competitive pressure: you need the Eureka moment as well. You could see less marginal improvements in the sense that they may have to stop labeling every new model something different if there is no real change, and trying to sell us new motherboards and new everything as well (except for a few details, imho Skylake  -> Kaby Lake is like AMD calling the FX-8370 something else than Piledriver, because that's just the 8350...). Competition will surely put more pressure on the marketing side of things (not just advertising, I mean all aspects of marketing: pricing, product design, choice of market segmentation, etc). Engineering-wise... wishing isn't doing  

 

 

[DELTAprime]

Not sure if this is a good number compared to other 4 cores but anyway...

[Tam3n]

Gave it a try on my ol' SB-E 24/7 clocks... Actually i7 3970X I found used on ebay is on it's way I should be good for a while longer on this setup then.

 

Here's gpu rendering (2x R9 290, OpenCL) results for comparison.

[Phate.exe]

2:04, 150 samples

Phenom II X6 1090T @ 3.6GHz on all cores, turbo core will send 3 of them to 4.1GHz, which is about as much as this motherboard (M4A88T-m, 4 phase power) will put up with and be gaming-stable.

 

Yeah, Zen will be a nice upgrade for me in a couple months.

[Mr.Dingle]

i5-4690k @ 4,4Ghz , 150 samples , 1:26.01

[dizmo]

At least I wasn't the worst

 

[Festive]

6700k @4.5ghz

 

https://gyazo.com/7d1a1915876d6a164e7ddcb0ccc7e39c

[siriq]

150 samples AMD FX 8350@4.8

01:07:03

[riddler0815]

27.78 seconds with dual E5-2670

[dexT]

Same as before, 4.5 on air. This time on Mint