Apple M1 Max’s GPU set to surpass all Intel laptops – MacGeneration

The first MacBook Pro presented at the beginning of the week are starting to be used in nature and we thus find results of performance measurements, in particular on the Geekbench site. The CPU had already been crossed at the beginning of the week, but it still lacked the integrated GPU in the Apple M1 Pro and Max. MacRumors To put my hand on a Metal test which measures the performance of the GPU, but it is not enough to draw all the useful conclusions.

The score displayed by this test is 68,870 points. As with all benchs, the number does not mean anything in itself and it is only interesting to compare machines with each other. So let’s compare: this score is more than three times higher than that obtained by the GPU of the Apple M1, which revolved around 22,000 points at best. It was expected, but this chip must compete with dedicated GPUs for laptops, this is the goal given by the manufacturer.

If we stick to Macs, this score is significantly better than that obtained by AMD’s Radeon Pro 5600M, which was the largest graphics card that could be installed in a 16-inch Intel MacBook Pro. She posted a result on the same test of about 42,500 points, the Apple M1 Max would do roughly 1.5 times better than it after this first test.

We would like to be able to use this measurement to compare with a portable RTX 3080, as Apple did during its keynote, but there are several obstacles to carrying out this comparison. The first is that Metal is an in-house API from Apple that does not exist on Windows. However, there is another bench of the MacBook Pro made with OpenCL, a standard that is also available on Windows.

The score of 60,167 points could seem disappointing in the face of Apple’s promises, since it is still lower to an RTX 3060, not to mention the high-end version cited by Apple. But remember that the OpenCL test is not optimized for Apple Silicon chips, which puts the MacBook Pro at a disadvantage. Moreover, the first GFXBench tests are also appeared in its database and these more concrete measures and close to a real gaming environment give a completely different picture.

This time, the Apple M1 Max stands out and reaches a level comparable to the scores of a RTX 3080 portable. From test to test, Apple’s chip may be slightly below Nvidia’s GPU, but the two are comparable and far ahead of Radeon Pro 5600M, which was the best option in a portable Mac. By the way, the Apple M1 Pro is also listed, with scores this higher than Nvidia’s GPU, which is a great surprise.

Another problem to note, we do not know which configurations of the Apple chips were tested each time. As we detailed yesterday, there are several variants of each chip, with for the Apple M1 Max two options for the number of GPU cores: 24 on the entry level, 32 on the high end.

Geekbench’s Metal test doesn’t say anything about the number of cores, the one with OpenCL in list 32, but go see if it’s the same Mac or even if it’s a reliable counter. As for GFXBench, this is not information that is given. If we are to believe Apple, the GPU of the Apple M1 Max should be four times faster than that of the Apple M1, which would favor the hypothesis of a flanged chip in the Metal score.

Let us remain cautious with these first measures, even if a first conclusion is undoubtedly emerging. The new MacBook Pro should be able to deliver the best graphics performance ever offered on a portable Mac. If it is the high-end version that has been tested, the score is much higher than that of the previous best option. And if it is “only” the variant with 24 cores, that means that we can expect even better results.

We can also measure the gap with GPUs dedicated to desktop computers. The best GPU currently supplied with the Mac Pro, the Radeon Pro W6900X, can thus achieve a Geekbench score with Metal which exceeds 182,000 points, more than 2.6 times that of the Apple M1 Max. Apple would have found the solution with its own chips, however: to multiply the hearts. If thewe believe the rumors, the Mac Pro could have four times as many GPU cores, which should be enough to regain the advantage.