- Mar 3, 2017
- 1,747
- 6,598
- 136
That's for data, max I've seen in nop chains is 17-18 bytes a cycle for a single core.
And since Z5 makes no performance-relevant LLC changes...
...well it shouldn't happen.
This is what they showed at ZEN4 X3D launch:View attachment 101647
14% faster than 13900K on average. Reality was like 4% faster.
Considering an AMD rep also said it won't be faster than Zen 4X3D, it's unlikely to be much, if at all, faster than the 14900k.
I think you've mixed up your 13900k and you're 14900k? 4% is what the gap roughly is, with 7800x3D that little bit aheadThis is what they showed at ZEN4 X3D launch:View attachment 101647
14% faster than 13900K on average. Reality was like 4% faster.
With the cherry-picked set of games and "Intel Baseline Profile", why not? )Unless AMD are outright lying the 9950X should be around 10-15% faster than the 14900K on average.
He also tests Strix while those tests on the slide are for Granite Ridge. There is known difference in L3 capacity, if his results are correct, there is difference in SIMD unit part of the core. Those could also explain stuff.This was clear enough already, but here's a little note regarding David Huang's results.
This was clear enough already, but here's a little note regarding David Huang's results.
With the cherry-picked set of games and "Intel Baseline Profile", why not? )
Just wondering if reviewers gonna use it with Intel CPUs in Zen 5 launch tests.
I think we can go off what AMD showed so far and that is 16% geo-mean. I expect that SIR 1T shows that on retail Granite Ridge parts. What Turin Dense can get on a different platform is a different topic.View attachment 101661
So the +32% SIR2017 cartel is back?
It's not an issue I think. What they were proud of is having the ISA parity between performance and compact cores, that remains true.The most interesting thing for me is that ZEN5 vs ZEN5 has more differences in terms of yArch than ZEN4 vs ZEN4c. Not a good move imo, they were so proud about the fact that ZEN4c is the same yArch, they had slides specifically for that. And now Strix has half the FPU without any official sign?
And the die area difference between Zen4 va Zen4C was bigger than the die area difference between Zen5 vs Zen5CThe most interesting thing for me is that ZEN5 vs ZEN5 has more differences in terms of yArch than ZEN4 vs ZEN4c. Not a good move imo, they were so proud about the fact that ZEN4c is the same yArch, they had slides specifically for that. And now Strix has half the FPU without any official sign?
And the die area difference between Zen4 va Zen4C was bigger than the die area difference between Zen5 vs Zen5C
Reviewers have so many options:With the cherry-picked set of games and "Intel Baseline Profile", why not? )
Just wondering if reviewers gonna use it with Intel CPUs in Zen 5 launch tests.
Reviewers have so many options:
Number 1 would be easiest for the reviewers but doesn't leave much confidence in the quality of their reviews.
- They could use the figures from their previous 14900K reviews
- They could re-run everything with the latest drivers and original BIOS profile maybe without any of the various security updates.
- They could re-run everything with the latest drivers and using the new supposedly stable BIOS profile.
Number 2 would please Intel PR but again won't be a great review. If their review sample has suffered from silicon degradation in the meantime, I'm sure Intel PR can sample a replacement.
Number 3 might get them on the Intel PR blacklist but is the only fair way to do their reviews.
No wonder they will, given that they raised this issue, and ironically, seem to be alone in thisI asked Hardware Unboxed, and they said they will re-run everything from scratch.
You don't hop on the hype train expecting it to be accurate, you join for the excitement and following disappointmentThis is a lesson I learnt, don’t catch the AMD hype bullet train. It’s especially prevalent in this site. From now on I’m trusting leaks that are less than a week away from announcement or release. It’s better that way.
Even when it releases we need to wait for third-party benchmarks since no company provides the full truth.
, the hype train hops on you.You don't hop on the hype train
Perhaps this explains the unusually long gap between announcement and release date.View attachment 101661
So the +32% SIR2017 cartel is back?
Strix launches on July 15, 42 days after announcement.Perhaps this explains the unusually long gap between announcement and release date.