- Mar 3, 2017
- 1,684
- 6,227
- 136
Maybe it’s time to put this argument to rest? In the last 4 years Apple had more ST gains than any x86 designer, and their rate of improvement actually accelerated with M4. Yes, some of those were achieved by increasing the power consumption (just like anyone else does). There is no evidence whatsoever that they are stalled or hurting for loss of talent. I don’t think that “they failed to increase Clang subtest IPC since 2021” is a good argument. Their IPC there is already ridiculously high. They might be at a practical limit.
my goodness. The level of defence for AMD. ARM did 16% YoY. 16% in 2 years and a "new" core is not good. Its average.The 16% IPC increase is a pretty good result. My biggest concern is that it took almost two years to achieve that.
If we take Apple, for comparison, M4 has 8 - 12% better IPC than M1 (in SPEC17 and GeekBench 6 without SME). Yes, during the last 4 years, Apple has had a smaller IPC increase than AMD in 2 years. Most of the performance increase comes from higher clock speeds.
I think it's because some "leakers" and "insiders" have been selling that this is going to be totally revolutionary, and clearly it's not. It's not terrible, but it's not ground-breaking.It's bad because AMD delivered moar 1t than that each core.
Pretty simple!
No, it's mid.I think it's because some "leakers" and "insiders" have been selling that this is going to be totally revolutionary, and clearly it's not. It's not terrible, but it's not ground-breaking.
real world performance is always better than synthetic and potentially manipulated benchmarksWe have results for SPEC for Apple chips and OpenBenchmarking is a joke.
SPECint isn't really synthetic.real world performance is always better than synthetic and potentially manipulated benchmarks
Nobody really noticed that they used 9950X and 7700X for Gen. Vs. Gen. comparison (GNR-03, probably IPC) ?
What the .... )
But Apple is increasing clock rates faster than AMD. AMD (and Intel) need to go for IPC because they can't get much frequency. Apple can still get frequency.
It’s Intel default settings per the endnotes.Interestingly, AMD is apparently comparing to 14900KS, not 14900K as indicated on the picture. And probably on old unstable Intel settings, not new "stable", but slower settings
Probably a typo in the footnotes, otherwise there would be no sense at all in the scores (no way a 16 core CU can be only +17% in a multi-core test compared to a 8 core CPU, and in the same slide notes all the tests are noted as n-threads). Not the first time slide decks are inconsistent.Nobody really noticed that they used 9950X and 7700X for Gen. Vs. Gen. comparison (GNR-03, probably IPC) ?
What the .... )
No I'm not thinking of userbenchmarkingreal world performance is always better than synthetic and potentially manipulated benchmarks
open bench is a joke because its real? man, we have user submitted benchmark threads here on the forums, those would easily be as legit... Are you thinking of userbenchmark? the site that is very not legit?
It’s Intel default settings per the endnotes.
Some of the metrics that are important for client are totally missing (7zip for example), the ones that are included WebXprt & Jetstream it’s 15% and 12% respectively. Also, why would they compare the 9950X v 7700X in the IPC chart? I’m willing to bet it does worse in 3rd party benchmarks than what AMD showed us tonight.
More importantly, who exactly is buying this thing over Zen 4 X3D? It doesn’t meaningfully beat RPL in 1T performance (either worse or at parity in javabloat and cinememe) and probably doesn’t outperform Zen 4 X3D in gaming.. so what is the market for SKUs like 9600X and 9700X?
Yes, but the power consumption of M4 has more than doubled compared to M1. One P-core in M4 running at 4.5 GHz consumes 7.2 to 9W. If you have 12 of them (in M4 Max), the total power consumption only for the CPU part will be nearly 100W.
Apple has to throttle them down back to M3 levels. As a result, the performance of M4 Max will be nearly the same as that of M3 Max on MT load.
If I hear the fans I'm going into low power mode. That's the biggest problem with x64 laptopsApple will simply kick up the fans in Macbook Pros
I'm sorry, but watts (more is better) is the meta and it's not the meta just in server.So far I've been able to count on Apple not doing that (unless I was actually doing something).
Oh yeah my P1 laptop is a pain. As soon as it's plugged in, fans turn onIf I hear the fans I'm going into low power mode. That's the biggest problem with x64 laptops
So far I've been able to count on Apple not doing that (unless I was actually doing something).
Good, so about the same as X Elite and Intel/AMD are much worse. Apple will simply kick up the fans in Macbook Pros and this is a non issue on the Studios. Well, this is the result of IPC stagnation..
What it is interesting is that no test used for the IPC increase over Zen4 is 1t. They are all multiple threads, of course there are lower threaded tests (like FC6 - which has a strange behaviour with AMD) and higher threaded tests.No, it's mid.
The rule of thumb is 25% 1t or more, or piss off.
They've been really, really consistent at not violating that, but alas.
Are you suggesting that I intentionally overhyped Zen5? Are you brain damaged?Then this is proof that you knew this entire time that the "30-40% IPC gain" claim was crock, and yet you still spread it. Given your previous record, I am not surprised.
balls in Intel's court now.As for Intel, I'm really excited to see the new E-cores (Skymont). If the leaked slides are correct and we will see a 38-60% IPC increase, the HX SKU (8 + 16) looks really nice. As for P-cores, I think we will see a moderate 15-20% IPC boost.
The problem isn't performance uplift, its timing:I think it's because some "leakers" and "insiders" have been selling that this is going to be totally revolutionary, and clearly it's not. It's not terrible, but it's not ground-breaking.
as a total platform, Strix Point is looking great.
It will be a great APU to get for a laptop.