Okay maybe oliterate wasn't the right word but some here and on Twitter had much higher expectations about Zen 5 performance.
The above tweet implied that x86 vendors had a core that would considerably beat Apple in terms of IPC, assuming thats what is meant by the use of "lapped". We now that Apple increased IPC on average 8% over M3 which was released 7 months ago, so that whole tweet about regarding Apple's IPC is wrong too. Not to mention the advantage Apple has in perf/w and the existing IPC lead also wasn't taken into consideration.
So in what way did Apple get lapped at Computex?
View attachment 100843
Look, I'm not pointing this solely on one person but I think the lesson we can take is we shouldn't jump take leakers words as truth and make false assumptions as to why the performance wasn't as expected. The best thing we can do now is wait and see the results for Zen 5.
Damn. Apple has the best ppw p-cores in the industry.. hope skymont can match the m2 or m3 in interger per wat performance
Okay maybe oliterate wasn't the right word but some here and on Twitter had much higher expectations about Zen 5 performance.
The above tweet implied that x86 vendors had a core that would considerably beat Apple in terms of IPC, assuming thats what is meant by the use of "lapped". We now that Apple increased IPC on average 8% over M3 which was released 7 months ago, so that whole tweet about regarding Apple's IPC is wrong too. Not to mention the advantage Apple has in perf/w and the existing IPC lead also wasn't taken into consideration.
So in what way did Apple get lapped at Computex?
View attachment 100843
Look, I'm not pointing this solely on one person but I think the lesson we can take is we shouldn't jump take leakers words as truth and make false assumptions as to why the performance wasn't as expected. The best thing we can do now is wait and see the results for Zen 5.
Pretty good. 1180 matches the top Snapdragon X Elite SKU.OK I’m scanning the thread and I’m not seeing any discussion of this. Am I missing something?
This is apparently from the Asus event at Computex giving a specific number for the 45 W Strix on Cinebench 2024.
1183 multithreaded.
Not bad, but I actually expected more. Looks like 7940HS is able to do 950+ with 54W, and with 50% more cores/threads I expected more than 25% uplift.This is apparently from the Asus event at Computex giving a specific number for the 45 W Strix on Cinebench 2024.
1183 multithreaded.
As long as it matches X Elite it is good enough.Not bad, but I actually expected more. Looks like 7940HS is able to do 950+ with 54W, and with 50% more cores/threads I expected more than 25% uplift.
Then drop the 7940HS to 45W, you'll get more than 25%.Not bad, but I actually expected more. Looks like 7940HS is able to do 950+ with 54W, and with 50% more cores/threads I expected more than 25% uplift.
Better perf/watt than Zen 4.A really interesting question I need answered is....
What is the power draw of Zen5?
Releasing a one off is one thing. Delivering regular, sizable performance increases is another. Apple has not delivered. Qualcomm has also struggled in this area. AMD has not. (no comment on Intel lol)Apple really impressed with M1. The improvments since have been more incremental. They do still have a great perf/W advantage which works great for battery life.
I am not Kepler and I don't follow the Twitterverse, nor do I claim to have inside knowledge. Keplers words aren't something I would've ever claimed but there will always be an exception. Sort of like the Zen 5 30+% IPC folks. Everything has to be taken with a grain of salt.
It is not easy to increase IPC gen to gen. That is, there is no magic bullet. AMD considers ~16% as generation improvement. For next couple generations, AMD will try to find bottleneck of Zen5 and will try to optimize.
We have to wait and see how Zen5 implements AVX512, if it has same negatives as Intel. Zen4 has balanced AVX512 implementation. AVX512 is not just about execution width, it also about addition very useful instructions.
As I mention previously, the next big opportunity for large IPC increase will come with introduction APX instruction set.
Personally, for desktop 65W-100W for CPU is ideal. Anything over should be extreme edition. The thing I don't like about 230W CPU is, it makes motherboard expensive for all the users because MB needs to be compatible with 230W CPU.
They'll be the same next gen!Not bad considering the benchmark doesn't favor Ryzens.
My 12700K does about 840 (with Intel XTU) DDR5-6000 CL28. Mobile Zen 5's ST score with possibly slower laptop RAM (most likely DDR5-5600. Can't access the leak source URL at work) is pretty awesome. It would take multiple generations in the past for mobile chips to catch up to desktop chips in performance but the gap is closing now at a faster rate.
Is there any reason to believe that the reviews will show higher IPC increase than the 16% that AMD announced?
Why would AMD be sandbagging Zen5 intentionally? Do they have a track record of doing that in the past? 🤔
It's good but still hard to judge........cuz mobile SKU has different TDP across different model though, here's some results collected from Jarrodstech:OK I’m scanning the thread and I’m not seeing any discussion of this. Am I missing something?
This is apparently from the Asus event at Computex giving a specific number for the 45 W Strix on Cinebench 2024.
1183 multithreaded.
| CB2024 | ST | MT |
| Lenovo Legion Slim 5 14" (2023) - Ryzen 7 7840HS (8c/16t) | 55 | 571 |
| ASUS TUF A16 (2023) - Ryzen 7 7735HS (8c/16t) | 92 | 676 |
| Framework Laptop 16 (2024) - Ryzen 7 7840HS (8c/16t) | 100 | 795 |
| ASUS Zephyrus G14 (2024) - Ryzen 9 8945HS (8c/16t) | 104 | 835 |
| Razer Blade 14 (2023) - Ryzen 9 7940HS (8c/16t) | 107 | 842 |
| Lenovo Legion Slim 5 16" (2023) - Ryzen 7 7840HS (8c/16t) | 106 | 889 |
| ASUS TUF A15 (2023) - Ryzen 9 7940HS (8c/16t) | 109 | 920 |
Was not aware i was asking anything for free. As i said, i paid significantly more (cca 700 vs cca 400) than before, for less features. I am not interested in excuses why they could not fit in this or that, cause it would be more expensive. It IS already more expensive.Then you understand that it doesn't come for free.. ? Is that what you're saying?
I am with B350 and 5950X . Another thing about my B350, it has 6 PCIe slots (I have utilized 4 of them). Most of the newer AM5 motherboards only have 3 slotsExpensive motherboards you say? (as I look despondently at the ASRock X399 Taichi setup)
Bloody thing is negated by a low tier B450 and used 3950X
Yeah, only comet in the nightskies would be clearer indicator than number of these accounts that the numbers presented by AMD are in fact bogus and perf increase is gonna be off the charts in reality :-DThere is another indicator that makes me optimistic about Zen5 real world performance.
The number of new accounts that have appeared in the last month that deride and polute the conversations taking place here on anandtech...
Poor souls wasting their time. How bored they must be coz their favorite brand launches their nextgen in October and then they have to spend something like $1000 to get new mobo and maybe a not very interesting Core Ultra 7 whereas existing AM5 users spend less than that and enjoy a 9950X. Really think Intel should swallow their pride and release ARL on LGA1700 with a socket adapter or something. In a month. To prevent a mass exodus of their gamer base jumping onto AM5.The number of new accounts that have appeared in the last month that deride and polute the conversations taking place here on anandtech...
It might score higher depending on what power limit the CPU is allowed. Since it seems like a 125W package, they could hypothetically allow the CPU up to 125W PPT when the GPU is idle.given 7945hx hits ~1700 MT CB24
strix halo should hit 2000+ no?
then strix point (1180) to halo is something like x2 difference
I'm curious how strong zen5c actually is compared to full zen5
Okay maybe oliterate wasn't the right word but some here and on Twitter had much higher expectations about Zen 5 performance.
The above tweet implied that x86 vendors had a core that would considerably beat Apple in terms of IPC, assuming thats what is meant by the use of "lapped". We now that Apple increased IPC on average 8% over M3 which was released 7 months ago, so that whole tweet about regarding Apple's IPC is wrong too. Not to mention the advantage Apple has in perf/w and the existing IPC lead also wasn't taken into consideration.
So in what way did Apple get lapped at Computex?
View attachment 100843
Look, I'm not pointing this solely on one person but I think the lesson we can take is we shouldn't jump take leakers words as truth and make false assumptions as to why the performance wasn't as expected. The best thing we can do now is wait and see the results for Zen 5.
The only integer IPC improvement they state is 7.3%.The M4 IPC improvement is still a bit unclear as Geekerwan showed 2 different results, one with 4.6% better integer IPC and one with 7.3% better. Not sure which one is more accurate,