Commentaires : Apple : le GPU du M1 Max est 3 fois plus rapide que celui du M1, bench à l'appui!

MDR comparé un Mac pro avec un PC portable doté d’une 3050 ti, le grand canyon est au niveaux du prix (X3), faut rendre à césar ce qui est à césar, « poire » resteras toujours « poire ».

1 « J'aime »

Les jeux ne sortaient déjà que rarement sous Mac Os, je ne suis pas sûr que le passage aux processeurs M1 arrange les choses… :sweat_smile:

Oui, il n’y a plus de CPU vraiment CISC.
je pense que l’avantage qu’a le M1 et qui explique le plus ses performances est l’inclusion de la mémoire dans le SOC : lignes de bus plus courtes égale moins de latence, et il est possible qu’elle fonctionne à des fréquences plus élevées que des barettes externes. Et en plus la mémoire sera mieux refroidie, utilisant le système de refroidissement du CPU.

Anandteh a benché le M1Max contre M1 i9-11980HK et Ryzen 5980HS : le M1 Max a une grosse avance dans les benchs nécessitant beaucoup de mémoire.
Possible qu’à l’avenir, la mémoire soit intégrée au CPU sur PC aussi vu ce que ça semble apporter en performances.

(page 1) Apple also doubles up on the memory interfaces, using a whopping 512-bit wide LPDDR5 memory subsystem – unheard of in an SoC and even rare amongst historical discrete GPU designs. This gives the chip a massive 408GB/s of bandwidth – how this bandwidth is accessible to the various IP blocks on the chip is one of the things we’ll be investigating today.

One large feature of both chips is their much-increased memory bandwidth and interfaces – the M1 Pro features 256-bit LPDDR5 memory at 6400MT/s speeds, corresponding to 204GB/s bandwidth. This is significantly higher than the M1 at 68GB/s, and also generally higher than competitor laptop platforms which still rely on 128-bit interfaces.

(page 2) From a single core perspective, meaning from a single software thread, things are quite impressive for the chip, as it’s able to stress the memory fabric to up to 102GB/s. This is extremely impressive and outperforms any other design in the industry by multiple factors, we had already noted that the M1 chip was able to fully saturate its memory bandwidth with a single core and that the bottleneck had been on the DRAM itself. On the M1 Max, it seems that we’re hitting the limit of what a core can do – or more precisely, a limit to what the CPU cluster can do.

(page 3)The workloads that put the most memory pressure and stress the DRAM the most, such as 503.bwaves, 519.lbm, 549.fotonik3d and 554.roms, have all multiple factors of performance advantages compared to the best Intel and AMD have to offer.

The performance differences here are just insane, and really showcase just how far ahead Apple’s memory subsystem is in its ability to allow the CPUs to scale to such degree in memory-bound workloads.

1 « J'aime »