![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
OMG look at this, the power draw, everyone jump ship now !!
5/9 charts are unidentifiable. you need to add links...I will reply to your personal remarks with a resume of main points of this discussion.
It started when posts #38 and #39 mentioned Steven's 'review' of the 7980XE. In post #43 I replied that I would wait for proper reviews, because Steven's recent reviews are pure garbage. I am not the only one that think this. Check what another user wrote in #53
I then gave a list of serious problems in Steven's recent 'reviews', including the fact that he used a motherboard with compatibility issues with the 7800X and damaged the chip during his 'review'. As demonstrated, AsRock released a BIOS fix for the 7800X about one month after Stevens published his 'review'. You can repeat the same point once again; it doesn't change anything. It is possible that Asrock marked the 7800X as incompatible before releasing the BIOS fix.
Now let us return to the original point. How wrong/biased and useless are Steven's reviews. We already have lots of reviews of the 7980XE. Stevens got the next power draw figures:
Other reviews got
i9-7980XE vs TR 1920X (other reviews): 105%, 108%, 98%, 109%, 100%, 79%, 108%, 89%.
i9-7980XE vs TR 1920X (Stevens): 157%.
i9-7960X vs TR 1950X (other reviews): 99%, 103%, 94%, 85%, 92%, 89%, 100%.
i9-7960X vs TR 1950X (Stevens): 125%.
So Stevens just got a much worse power consumption figures for Intel chips than other reviews. How unexpected!
Guess he better learn how to scale his charts.
I will reply to your personal remarks with a resume of main points of this discussion.
It started when posts #38 and #39 mentioned Steven's 'review' of the 7980XE. In post #43 I replied that I would wait for proper reviews, because Steven's recent reviews are pure garbage. I am not the only one that think this. Check what another user wrote in #53
I then gave a list of serious problems in Steven's recent 'reviews', including the fact that he used a motherboard with compatibility issues with the 7800X and damaged the chip during his 'review'. As demonstrated, AsRock released a BIOS fix for the 7800X about one month after Stevens published his 'review'.
You can repeat the same point once again; it doesn't change anything. It is possible that Asrock marked the 7800X as incompatible before releasing the BIOS fix.
Now let us return to the original point. How wrong/biased and useless are Steven's reviews. We already have lots of reviews of the 7980XE.
Other reviews got
<clip>
i9-7980XE vs TR 1920X (other reviews): 105%, 108%, 98%, 109%, 100%, 79%, 108%, 89%.
i9-7980XE vs TR 1920X (Stevens): 157%.
i9-7960X vs
i9-7960X vs TR 1950X (Stevens): 125%.
So Stevens just got a much worse power consumption figures for Intel chips than other reviews. How unexpected!
/jk I have to say, the only time I have ever had regrets is when I did not purchase what I wanted too in the first place. I have a 1700 and 1700X and am happy with them but, I would love me some Threadripper just because. 
I will reply to your personal remarks with a resume of main points of this discussion.
It started when posts #38 and #39 mentioned Steven's 'review' of the 7980XE. In post #43 I replied that I would wait for proper reviews, because Steven's recent reviews are pure garbage. I am not the only one that think this. Check what another user wrote in #53
I then gave a list of serious problems in Steven's recent 'reviews', including the fact that he used a motherboard with compatibility issues with the 7800X and damaged the chip during his 'review'. As demonstrated, AsRock released a BIOS fix for the 7800X about one month after Stevens published his 'review'. You can repeat the same point once again; it doesn't change anything. It is possible that Asrock marked the 7800X as incompatible before releasing the BIOS fix.
Now let us return to the original point. How wrong/biased and useless are Steven's reviews. We already have lots of reviews of the 7980XE. Stevens got the next power draw figures:
Other reviews got
i9-7980XE vs TR 1920X (other reviews): 105%, 108%, 98%, 109%, 100%, 79%, 108%, 89%.
i9-7980XE vs TR 1920X (Stevens): 157%.
i9-7960X vs TR 1950X (other reviews): 99%, 103%, 94%, 85%, 92%, 89%, 100%.
i9-7960X vs TR 1950X (Stevens): 125%.
So Stevens just got a much worse power consumption figures for Intel chips than other reviews. How unexpected!
. I hope Intel sells just tank to get them to be more price competitive.All the original points are in the SKL-X thread here
https://hardforum.com/threads/skylake-x-core-i9-lineup-specifications-and-reviews.1933735/
The information is scattered about several pages, but I don't remember what pages contains each point. If you read the whole thread you will find the information. Luckily I wrote on a disquss thread a resume of issues with one of his last reviews:
(i) The guy tested in GPU-bound and frame-limiting settings. That is the reason why overclocking the 7800X by a huge 34% did only bring 3% higher framerates (the performance of the SKL-X CPU was being crippled), and the reason why the OC 1600 is able to caught the 7800X.
When people asked him why overclocking the SKL-X chip didn't bring any relevant performance benefit. The guy tried to excuse himself pretending that was something wrong with the Skylake-X chip
but the problem wasn't in the SKL-X chip, but on his testing using GPU-bound and frame-limiting settings. That is the reason why overclocking the i7-7700k by 16% did only bring 2% higher framerates (the performance of Kabylake CPU was being crippled as well).
(ii) The guy finally admitted he didn't even test a retail 7800X chip, but he used an ES or a QS (he doesn't seem to grasp the difference between qualification sample and engineering sample)
(iii) If this wasn't enough, he used a motherboard was incompatible with the 7800X
http://www.asrock.com/MB/Intel/X299 Taichi/index.asp#BIOS
and he managed to burn the CPU thanks to that
In short, about everything what could be wrong was wrong in their recent reviews. Call it plain incompetence or call it bias. The point is that I don't trust those guys anymore and I better will wait to proper reviews before making my opinion about the new XE model.
(i) Price is not a linear function of performance.
(ii) Not everyone make a purchase decision in base to performance/price ratios. Your purchase of 1800X proves it.
(iii) Why to use only CPU pricing? Some people purchases complete systems. Others got CPU+mobo-RAM combos.
(iv) Is that 20--25% more performance an average on non-gaming, non-AVX workloads? Because I have been the i9-7980XE render images in 3DS max in 42% less time than TR-1950X, I have seen in the Intel Core i9-7980xe thread a game benchmark with the i9 giving 49% more FPS than TR. On AVX workloads the gap is higher
(v) The ten reviews given in this thread show that either the 7980XE consumes about the same power than the 1950X or even less. So the 7980XE wins in the performance/power ratio.
(vi) The X299 platform is more complete, faster, and flexible tan the X399 platform.
BS. You would have to cherry pick the hell out of some sites benchmarks to get TR to pull more than 7980X. AMD has shown to easily use less power in most all sites I have seen. The only sites to show lower usage on their i9 could not accurate speak to throttling.I mentioned in my post it was an AVX workload. and I said the performance gap has to be higher than in ordinary x86 workloads. I gave that AVX workload as example of the kind of performance than ordinary reviews aren't measuring when only test Blender, CineBench, Corona,...
7980X being only 2x TR performance and 7900X being only 2x faster than 1800X in that benchmark makes me suspect that AVX workload only uses 256bit vectors, because the performance gap would be higher with 512bit. When 512bit is enabled the 7900x can be about 3.5 times faster than R7 RyZen
AVX512 is a standard in HPC and servers. I hope it becomes more popular in desktops: renderers, and scientific/financial applications would benefit from wider vectors
http://www.sisoftware.eu/2016/02/24/future-performance-with-avx512-in-sandra-2016-sp1/
If one chip has better power consumption policies it consumes less power. So now this is a problem?
I gave above ARS power measurements for i9-7960X @4.1GHz and TR-1950X @3.9GHz. The Intel chip is more efficient. The TR chip hit a wall, but if it could be clocked higher and get Intel clock, then the total power consumption would be about 40W higher than for the Intel patform.
I also gave Kitguru power measurements for i9-7980XE @4.6GHz vs TR-1950X @3.95GHz. The AMD system consumed about the same power 515W, 520W than the i9: 510, 550W. Again if the TR chip could hit 4.6GHz would consume much more power than the i9.