Your constant obsession for all things AMD isn’t healthy mate. Nothing Shawn said was out of line. This could very well be AMD’s last chance to get things right, so when a benchmark gets released showing them trailing Intel’s previous generation architecture it isn’t unreasonable to voice concern.
Really? Are you ready?
The Zen ES reports eight cores, 16 threads and is a 14nm part yet it can’t keep pace with Intel’s old chips?
Cannot? Yes, it cannot because
@Shawn Knight cannot do anything else than just look at results.
I begin looking at
those results only. Estimated clock speeds during test:
i5-4670K: 3.5 GHz (52.6 FPS)
i7-4790K: 3.8 GHz (65.4 FPS)
Zen ES: 3.0 GHz (58 FPS)
FPS per 1 GHz:
i5-4670K: 15.0
i7-4790: 16.97
Zen ES: 19.3
So Zen's IPC is much higher than Intel CPU's!
If this is true, then Zen's IPC is much bigger than Haswell's.
Now, 8 cores cannot keep up with 4 cores?
i7-4790K: cores 4, threads 8, 3.4 GHz base, 3.8 GHz boost, 84W TDP (with no "integrated chipset")
Zen ES: cores 8, threads 16, 2.8 GHz base, 3.2 GHz boost, 95W TPD (with many chipset functions integrated)
CPU's have quite same TDP, AMD just have double number of cores.
Now,
assuming both CPU's have same IPC and clocks during test are 3.8 GHz for i7-4790 and 3.0 GHz for Zen. For 4 cores used, i7 is 8 percent faster. 5 cores used, i7-4790 is STILL 2% faster.
For Zen to surpass i7-4790 with same IPC that test needs to fully utilize at least 5.1 cores!
And so Zen sucks because it cannot keep pace when not all cores are used (n)
I understand that your job is to defend those who write at this site but as I clearly showed above, it seems that
@Shawn Knight have no ability to do even basic calculations based on results.
Shawn pointed out that the results might not be accurate/legitimate and he also said this is just a single benchmark so it’s tough to draw full conclusions.
And still AMD is bad because it cannot keep pace against quad cores on test that is not designed to test 8 core CPU's?
Of course looking at the graph the AMD Zen processor is clocked 20% slower which means the 11% decline in performance when compared to the Core i7-4790 is actually very good. Keep in mind I am at this point assuming Zen overclocks well and not like a GCN 4th gen GPU.
Exactly, you finally managed to do some simple calculations, not just looking what CPU has bigger number (y) Something
@Shawn Knight didn't do.
Another blunder here:
If the retail product is pushing speeds closer to 4GHz, performance will no doubt improve significantly.
Again, some simple knowledge tells this is unlike to happen, comparable AMD vs Intel:
Zen ES: cores 8, threads 16, 2.8 GHz base, 3.2 GHz boost, 95W TPD
Xeon E5-2640 v3: cores 8, threads 16,
2.6 GHz base,
3.4 GHz boost, 90W TDP
As even Intel cannot make octa core even close 4 GHz with same TDP, not very likely can AMD. Intel needs 140W TDP for that 4 GHz.
8-core Zen is designed to give good IPC and high core count at low power consumption and price. Something has to suffer, and that's clock speed.
Not only that but utilization isn’t that important here, as long as both processors offer the maximum number of cores the game can utilize then you get a very clear picture of IPC performance.
Indeed, did you see any IPC analysis on "article"? I didn't. So actually even you admit that "article" was mostly BS.
It is probably easy for you to overlook the fact that these results have been recorded using an RX 480 which is without question limiting the performance of the i7-4790. I personally doubt the legitimacy of these findings given the sources credibility but if they are true then for me they leave more questions than answers.
Exactly, that test don't tell much but
@Shawn Knight makes direct assumptions based on those numbers. Then those assumptions should be something better than "Intel 65.4, AMD 58.0, AMD sucks."
@Shawn Knight and
@Steve : feel free to make that article better using anything I wrote on this post.