No, i know they ran it stock. I reference the users pushing these to 4Ghz and up.Seems you think they were doing O/C'ing, no, that wasn't the case.
Wait wait wait, you say that my claim that evidently game optimizations allow Ryzen to keep up with BDW-E in games as well is erroneous? The one thing i did not expect, i admit.That data in no way allows you to draw that (erroneous) conclusion.
It does not stop 4Ghz+ Ryzen from being a furnace akin to 4.3-4.4Ghz 6900k. 3.3-3.5Ghz Ryzen? Oh yeah, it is efficient. 4Ghz? Not a chance.But Ryzen is great in terms of efficiency.
Wake up dude. You still dont get the memo dont you?
You are referring to the handbrake test.
Did you notice the performance difference here for the same test?
A 1700 non x beats a 6900. Read again.
A 1700x beats a 6950. Yes the 350 usd cpu beats the 10 core 1800 usd intel. Read again.
Blimey, 1100$ 6900k is only 16% ahead of a 500$ 1800x with bugged bios even when using its full AVX2 and quad channel power, thats actually MUCH better than i imagined, i thought AVX2 would see a 50% beat down in favour of 6900k with its full instructions.Talk about complete hyperbole and lack of proper analysis. Yes, Ryzen's performance is exceptional, especially given its low price relative to the competition. However, lets get some real perspective.
First off, the Hexus review (which was God awful) used an old version of Handbrake, version 0.10.5. At the time of Ryzen's release, Handbrake was up to 1.02. This makes a big difference, because 1.00 was the milestone version and it has lots more enhancements and refinements compared to previous versions.
Secondly, x.264 is an old codec which does not respond well to multithreading and SIMD optimization generally speaking. x.265 is much better in that regard being a modern codec, and can take advantage of multithreaded CPUs and advanced SIMD instructions much more effectively than x.264.
Observe. Under H.264, Ryzen is leading the 6900K slightly, and the FX-9590 and Phenom II X 6 1090t don't look bad at all. But when you look at the H.265 scores, you see that the 6900K is ahead of the 1800x by 16%, and that the 6850K is almost neck and neck with the 1800x. The FX-9590 and the Phenom II X6 are blown out of the water and utterly decimated respectively. The reason why this is, is because the Core i7 CPUs gains significantly more performance than the Ryzen CPUs from AVX2 optimization (which H.265 supports heavily), and also because H.265 also uses more CPU cores much more effectively than its predecessor.
Presumably, longer tests would result in Intel gaining even larger leads over Ryzen.
So the point is, be careful about coming to conclusions on these handbrake benchmarks, because there are many factors to consider. Ryzen might look more competitive on older codecs like X.264 because they don't make effective use of modern CPU capabilities. You can expect AV1 to have excellent support or multithreading and SIMD optimizations like x.265, but hopefully even faster and higher quality. So in the long run, Intel CPUs are still going to beat Ryzen due to their more formidable SIMD capabilities if you ask me.
No, i know they ran it stock. I reference the users pushing these to 4Ghz and up.
Wait wait wait, you say that my claim that evidently game optimizations allow Ryzen to keep up with BDW-E in games as well is erroneous? The one thing i did not expect, i admit.
It does not stop 4Ghz+ Ryzen from being a furnace akin to 4.3-4.4Ghz 6900k. 3.3-3.5Ghz Ryzen? Oh yeah, it is efficient. 4Ghz? Not a chance.
What? Its obvious what you were suggesting, as you were replying to a software update post, who mentioned overclocking?No, i know they ran it stock. I reference the users pushing these to 4Ghz and up.
Wait wait wait, you say that my claim that evidently game optimizations allow Ryzen to keep up with BDW-E in games as well is erroneous? The one thing i did not expect, i admit.
It does not stop 4Ghz+ Ryzen from being a furnace akin to 4.3-4.4Ghz 6900k. 3.3-3.5Ghz Ryzen? Oh yeah, it is efficient. 4Ghz? Not a chance.
Blimey, 1100$ 6900k is only 16% ahead of a 500$ 1800x with bugged bios even when using its full AVX2 and quad channel power, thats actually MUCH better than i imagined, i thought AVX2 would see a 50% beat down in favour of 6900k with its full instructions.
Still you do raise a good point, newest software which uses full instructions must be used for comparisons.
Well, i suppose that's the first time in 10 or so years when someone actually adding context leads to misunderstanding.What? Its obvious what you were suggesting, as you were replying to a software update post, who mentioned overclocking?
That I did not make. So, how is it obvious for me to understand you were addressing a claim i did not make.It's obvious I was addressing your claim that Ryzen is a "furnace".
Folks, this is how alternative reality looks like: a knowledgeable poster intentionally trying to deceive the forum that a CPU consuming a bit under 150W @ 3.8Ghz is a furnace when compared with another using 150W @ 4Ghz. Yes, you read that right, they use more or less the same power for more or less the same frequency, but the AMD product is a furnace.It does not stop 4Ghz+ Ryzen from being a furnace akin to 4.3-4.4Ghz 6900k. 3.3-3.5Ghz Ryzen? Oh yeah, it is efficient. 4Ghz? Not a chance.
@Carfax83
Interesting, but does anyone have power consumption figures when running H.265 and the AVX2 code path? Because they are bound to be totally different the H.264 ones then.
You are absolutely correct, 4ghz 6900k in Prime95/Linpack/Other burn test is a furnace as well. In fact, stock 6900k in those is a furnace. Does that translate to other well-threaded apps that are not created to turn a CPU into a furnace? No, last time i checked.Folks, this is how alternative reality looks like: a knowledgeable poster intentionally trying to deceive the forum that a CPU consuming a bit under 150W @ 3.8Ghz is a furnace when compared with another using 150W @ 4Ghz.
Also using launch control too often with a Tesla significantly reduces reach, and increases overall risk. Similar to Ryzen OC. So you have a point, and I added some context again. /sWell, i suppose that's the first time in 10 or so years when someone actually adding context leads to misunderstanding.
No, i was suggesting only 1 thing: Ryzen at 4Ghz+ to which some users push it is a furnace. Want to deny it? You could try. What they do it for? For 10% over ~3.6Ghz.
That I did not make. So, how is it obvious for me to understand you were addressing a claim i did not make.
Those well threaded apps not created to turn a CPU into a furnace will have a similarly limited impact on 1800X as well. In fact, even the 4.4Ghz 6900K will not be a furnace when using anything but a burn test, see the gaming run with 90W average power consumption. Funny enough, this also invalidates your argument, not only for 1800X, but for 6900K as well.You are absolutely correct, 4ghz 6900k in Prime95/Linpack/Other burn test is a furnace as well. In fact, stock 6900k in those is a furnace. Does that translate to other well-threaded apps that are not created to turn a CPU into a furnace? No, last time i checked.
By liking your comment it seems he misunderstood your sarcasm.Also using launch control too often with a Tesla significantly reduces reach, and increases overall risk. Similar to Ryzen OC. So you have a point, and I added some context again. /s
H.264 is by far the most widespread codec in use today. It will remain so in the next couple of years, especially if the applications where h.265 really shines, ie. 4K, are restricted to particular CPUs and GPUs with FFHs.Exactly. H.264 is on the way out, and using it to determine a modern CPUs processing power isn't too bright.
Brave generalization, when your own charts disprove validity of it, don't you think? Why don't i try my hand at it too? We see from Ryzen chart that relatively cool (on the CPU) Blender consumes about 70% more power than gaming run on Ryzen. Extrapolating it to 4.4Ghz 6900k we land with firmly furnace level 145W of power consumption on CPU (+presumably VRM). Now, furnace is subjective qualifier, and the whole thing, as Dresdenboy neatly put it is me talking about utterly irrelevant things, but heh, it went funnier than i had expected.In fact, even the 4.4Ghz 6900K will not be a furnace when using anything but a burn test, see the gaming run with 90W average power consumption.
No, i understood it perfectly, and it was the better rebuttal of my post than any of you had managed. I appreciate wit.By liking your comment it seems he misunderstood your sarcasm.
Performance segments Intel has created to maximize their profits. Ryzen is showing that you can have both. You can have R5/R3 or the upcoming x399 16/12c processors but you could very well sit in the middle with R7. AMD is mixing the pot and in a very good way from consumer perspective.There is no way Intel can justify their prices or every product segment in the future.If we look at the stilt fmax vs V graph there is a liniar correlation all the way up to 3.3GHz. I stlill dont understand it is so high?
Obviously more die mm2 solves a lot of efficiency problems if its prioritized. Also showed for 7700 vs 1700. But we are looking at this as consumers and 6950 is hardly comparable to 1800 or 1700 even if it in this specific workload appears to be the same. Its different price, dev and process cost and - for the most part - performance segments. How would a x399 16c perform for perf and efficiency?
Thats the product that can be compared to a 6950 class products (more likely the hedt skl 12c derivative)
So now you consider 140W to be furnace level power consumption, and you consider extrapolating gaming to Blender power numbers between 2 different scenarios to be fair game. Fine by me, let's play!Brave generalization, when your own charts disprove validity of it, don't you think? Why don't i try my hand at it too? We see from Ryzen chart that relatively cool (on the CPU) Blender consumes about 70% more power than gaming run on Ryzen. Extrapolating it to 4.4Ghz 6900k we land with firmly furnace level 145W of power consumption on CPU (+presumably VRM).
To the ones you are saying that there are minimal Intel game optimizations on the market quote from Oxide: "For basically 5 years”, I was told, Oxide and other developers have dedicated their time to “instruction traces and analysis to maximize Intel performance” which helps to eliminate poor instruction setup. So first Ryzen game optimization cames and gives us ~20% more performance. And as fast this update was announced it seems that the job wasn't even a complicated to achieve. Gives bright view to Ryzen's future game release performance.What happens when games are tuned for Ryzen?
https://www.pcper.com/reviews/Processors/Ashes-Singularity-Gets-Ryzen-Performance-Update
Yes a 6850 is equal to a 1800x in avx2 workload for eg h265. We have an entire thread with results showing so. Its nearly best case perf for Intel (best would have been more branched code) and its still similar cost. So from a perf/ $ Intel solution still looks meh.Talk about complete hyperbole and lack of proper analysis. Yes, Ryzen's performance is exceptional, especially given its low price relative to the competition. However, lets get some real perspective.
First off, the Hexus review (which was God awful) used an old version of Handbrake, version 0.10.5. At the time of Ryzen's release, Handbrake was up to 1.02. This makes a big difference, because 1.00 was the milestone version and it has lots more enhancements and refinements compared to previous versions.
Secondly, x.264 is an old codec which does not respond well to multithreading and SIMD optimization generally speaking. x.265 is much better in that regard being a modern codec, and can take advantage of multithreaded CPUs and advanced SIMD instructions much more effectively than x.264.
Observe. Under H.264, Ryzen is leading the 6900K slightly, and the FX-9590 and Phenom II X 6 1090t don't look bad at all. But when you look at the H.265 scores, you see that the 6900K is ahead of the 1800x by 16%, and that the 6850K is almost neck and neck with the 1800x. The FX-9590 and the Phenom II X6 are blown out of the water and utterly decimated respectively. The reason why this is, is because the Core i7 CPUs gains significantly more performance than the Ryzen CPUs from AVX2 optimization (which H.265 supports heavily), and also because H.265 also uses more CPU cores much more effectively than its predecessor.
Presumably, longer tests would result in Intel gaining even larger leads over Ryzen.
So the point is, be careful about coming to conclusions on these handbrake benchmarks, because there are many factors to consider. Ryzen might look more competitive on older codecs like X.264 because they don't make effective use of modern CPU capabilities. You can expect AV1 to have excellent support or multithreading and SIMD optimizations like x.265, but hopefully even faster and higher quality. So in the long run, Intel CPUs are still going to beat Ryzen due to their more formidable SIMD capabilities if you ask me.
Just like you consider gaming run a valid example of well-threaded work. Anything is fair game at this point. And yes, i do consider 140W on CPU "furnace" level, because i use my PC that consumes about 160W at the wall under gaming load for heating myself in winter.So now you consider 140W to be furnace level power consumption, and you consider extrapolating gaming to Blender power numbers between 2 different scenarios to be fair game. Fine by me, let's play!
1800X @ stock uses 95W in Blender, and the Luxrender gives us power usage scaling from stock to oc (+27%), hence 1800X @ 3.8Ghz in Blender would use ~120W.
No, it does not unlock furnace achivement, i said as much.3.8Ghz starts to meddle into furnace territory.
H.264 will be relevant for many years to come, but you're right that H.265 is slowly taking more role in this area. Also the question is always about the money, what you get per buck. Intel's HEDT line is suddenly now way too expensive thanks to AMD. Also not sure how much the full AVX2 is actually required for H.265 or is there a way improve Ryzen's performance here like Oxide just made with AoS. Maybe not, but just wondering.Well I already acknowledged that Ryzen has fantastic price/performance, and that AMD did an amazing job of improving over their previous offerings; which to be honest wasn't too difficult as their previous offerings stank. But lets not pretend it's all roses for Ryzen.
That sample benchmark I pulled from Computerbase.de is very small as far as real world encoding goes. Reviewers typically choose very small samples with the obvious intent being to save time. But, the trend is clear in the sense that a longer and more realistic encoding would have Intel increase its margin of victory even more.
The thing is, that Intel not only has the full 256 bit registers for AVX2, it also has much greater bandwidth on tap so it can sustain those AVX/AVX2/FMA operations much better than Ryzen can I believe.
Exactly. H.264 is on the way out, and using it to determine a modern CPUs processing power isn't too bright. H.265 and the upcoming AV1 will be the real tests.
This Ryzen Power Consumption Graph is quite informative, but the truth is Ryzen is YET to be torture tested. This is very curious to me. That P95 result is a farce.Folks, this is how alternative reality looks like: a knowledgeable poster intentionally trying to deceive the forum that a CPU consuming a bit under 150W @ 3.8Ghz is a furnace when compared with another using 150W @ 4Ghz. Yes, you read that right, they use more or less the same power for more or less the same frequency, but the AMD product is a furnace.