- Mar 3, 2017
- 1,680
- 6,185
- 136
Thanks, but I truly can't say I am happy with this.June 2024, that MS AI PC event.
You're welcome.
Sorry, I am "CB rules them all", although with the release of CB24 not sure any more considering M3 saw a big increase compared to CB R23.
And I think It's a good question, If those M3 scores are not also thanks to the compiler doing It's magic.
Won't say CB is the best, I just consider It better than Geekbench.So you're saying you believe CB is the best benchmark, unless it makes Apple look too good in which case you'll decide to believe in some other benchmark that shows you what you want to see?
OK then, so its 30+ percent or 10+ percent? Cause, while not this SIR, but CB24, the rumored results pointed toward that lower number. Or could SIR bench results and CB scores be so vastly different?Applebench isn't relevant, it's all about SIR2017.
Won't say CB is the best, I just consider It better than Geekbench.
And It's true that Apple SoCs now perform a lot better in CB24 vs competition than they did in the previous version. Not sure what was changed.
Whomstthe rumored results pointed toward that lower number
I was pretty clear about that eh.OK then, so its 30+ percent or 10+ percent?
CB results were always lower for Ryzen versus the real average IPC:OK then, so its 30+ percent or 10+ percent? Cause, while not this SIR, but CB24, the rumored results pointed toward that lower number. Or could SIR bench results and CB scores be so vastly different?
Thought it was GB 5 that tracks pretty well with SPEC? Maybe GB 6 ST does too, but I doubt GB 6 MT does. IIRC there was a paper on it, but idk if that was just a fever dream or what lolThat GB6 1T also tracks with both SPEC 1T and browser performance (the one workload to rule them all) is just cherry on top for us ultra-reductionists
I feel like the pattern is much more complicated than just "AMD does worse on CB" and more to do with the specific arch implementation. Bcuz Zen 5 is such a large change, I have no reason to believe that pattern would hold.CB results were always lower for Ryzen versus the real average IPC:
Zen 3 (avg IPC uplift ~19%; CB R20 1T IPC uplift 13%):
View attachment 90374
Zen 4 ( avg IPC uplift 13%; CB R23 1T uplift 9%):
View attachment 90375
For both Zen 3 and Zen 4 the ratio is roughly 0.68 between the two. If this continues to be the case, and if Zen 5 has ~30% avg IPC uplift (geomean, just like shown above), it is possible that CB 1T could be ~20% better vs Zen 4. Given how much bigger the core will be, and what was changed, getting 2x the IPC uplift of what Zen 4 got is not unreal at all.
PS The above Zen 5 numbers are pure speculation by me.
AMD first gets SPEC benchmark percentage uplift, then they collect a list benchmarks that would average to SPEC percentage increase. I think that is how AMD is coming up with these charts. I remember Ian or some other reviewer mentioning this.CB results were always lower for Ryzen versus the real average IPC:
Zen 3 (avg IPC uplift ~19%; CB R20 1T IPC uplift 13%):
View attachment 90374
Zen 4 ( avg IPC uplift 13%; CB R23 1T uplift 9%):
View attachment 90375
For both Zen 3 and Zen 4 the ratio is roughly 0.68 between the two. If this continues to be the case, and if Zen 5 has ~30% avg IPC uplift (geomean, just like shown above), it is possible that CB 1T could be ~20% better vs Zen 4. Given how much bigger the core will be, and what was changed, getting 2x the IPC uplift of what Zen 4 got is not unreal at all.
PS The above Zen 5 numbers are pure speculation by me.
Thought it was GB 5 that tracks pretty well with SPEC? Maybe GB 6 ST does too, but I doubt GB 6 MT does. IIRC there was a paper on it, but idk if that was just a fever dream or what lol
I feel like the pattern is much more complicated than just "AMD does worse on CB" and more to do with the specific arch implementation. Bcuz Zen 5 is such a large change, I have no reason to believe that pattern would hold.
Because some people who don't know better will use it to compare CPUs with different number of cores. GB MT score is useless for that. "Oh my god M3 Max beats M2 ultra with just half the cores!!!"I don't get why people think GB6 MT not scaling perfectly is a problem.
that was true for Zen 3 but for Zen 4 AMD claimed 13% when it was 8-9% in SPEC ST.AMD first gets SPEC benchmark percentage uplift, then they collect a list benchmarks that would average to SPEC percentage increase. I think that is how AMD is coming up with these charts. I remember Ian or some other reviewer mentioning this.
The same can be said in reverse, we've been fooling people for years into believing the high core count CPUs are twice or three times as fast in consumer MT workloads.Not scaling perfectly isn't the problem, calling it "MT" is (for the reason @SteinFG mentioned).
It depends what are you actually trying to measure. If you are measuring single instance "of problem solving" aka consumer pushes the "enter" button and how long coffee break should he takeI don't get why people think GB6 MT not scaling perfectly is a problem. I don't know what it is doing, maybe it is being stupid, but in the real world most problems aren't like SPEC or CB's MT where the way they do it allows perfect scaling until you hit system limits like memory bandwidth or cache coherence traffic. In the real world many MT problems have the various threads cooperating with each other where one depends on the results from another and so on, and sometimes threads will have to pause and wait until other threads whose results they depend on are complete.
What type of benchmark you look at should have some relationship to the type of software you run. If your software is of that "cooperative" nature, looking at MT results based on embarrassingly parallel workloads isn't going to tell you much.
Because some people who don't know better will use it to compare CPUs with different number of cores. GB MT score is useless for that. "Oh my god M3 Max beats M2 ultra with just half the cores!!!"
And they are the most significant part of the market.People who don't know better will do a lot of things. At least this way they are easy to identify.
I agree, care to chime in Arston?that was true for Zen 3 but for Zen 4 AMD claimed 13% when it was 8-9% in SPEC ST.
no such thingDoes anybody know what upgrades or differences the B750/X770 motherboards will bring with Zen 5?
There is no new chipset, if B750/X770 exist it will just be a refresh of current mobos.Does anybody know what upgrades or differences the B750/X770 motherboards will bring with Zen 5?