Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple GPUs speculation: TBDR, RT and the benefit of wide ALUs
- Thread starter leman
- Start date
- Sort by reaction score
So it seems, but we will need to wait for more detailed developer docs. I wouldn't expect RT to be priority for this release though, we only got some entry-level hardware. You won't be running RT games on that anyway.
They are hiding a lot of hardware details: clocks, RAM type, TDP... RAM will most likely be LPDDR4/5 with some sort of wide multi-channel configuration. Probably around 80GBps bandwidth, something that will be plenty for a chip of this spec.
Yeah, I think so. I'm pretty sure this is dual channel LPDDR4X. Compare the image you saw today to this image of the A12X:
Which uses LPDDR4X. I think, frankly, LPDDR5 supplies were more limited than they expected.
You can read the footnotes/fineprint on the GPU for performance comparisons:
"Testing conducted by Apple in October 2020 using preproduction 13-inch MacBook Pro systems with Apple M1 chip and 16GB of RAM using select industry-standard benchmarks. Comparison made against the highest-performing integrated GPUs for notebooks and desktops commercially available at the time of testing. Integrated GPU is defined as a GPU located on a monolithic silicon die along with a CPU and memory controller, behind a unified memory subsystem. Performance tests are conducted using specific computer systems and reflect the approximate performance of MacBook Pro."
The definition basically excludes Ryzen APU's since the Ryzen chiplet is on a separate silicon die from the CPU and memory controller
MacBook Pro
MacBook Pro laptop with M4, M4 Pro, and M4 Max chips. Built for Apple Intelligence. Up to 24 hours of battery life. Liquid Retina XDR display.
www.apple.com
"Testing conducted by Apple in October 2020 using preproduction 13-inch MacBook Pro systems with Apple M1 chip and 16GB of RAM using select industry-standard benchmarks. Comparison made against the highest-performing integrated GPUs for notebooks and desktops commercially available at the time of testing. Integrated GPU is defined as a GPU located on a monolithic silicon die along with a CPU and memory controller, behind a unified memory subsystem. Performance tests are conducted using specific computer systems and reflect the approximate performance of MacBook Pro."
The definition basically excludes Ryzen APU's since the Ryzen chiplet is on a separate silicon die from the CPU and memory controller
That should mean the comparison is against the 96 EU Tiger Lake part. Yet somehow I don't think that's what they used. An 8 core Apple GPU would probably squeak by a 96 EU Xe tuned to 28W, but it would be close.You can read the footnotes/fineprint on the GPU for performance comparisons:
MacBook Pro
MacBook Pro laptop with M4, M4 Pro, and M4 Max chips. Built for Apple Intelligence. Up to 24 hours of battery life. Liquid Retina XDR display.www.apple.com
"Testing conducted by Apple in October 2020 using preproduction 13-inch MacBook Pro systems with Apple M1 chip and 16GB of RAM using select industry-standard benchmarks. Comparison made against the highest-performing integrated GPUs for notebooks and desktops commercially available at the time of testing. Integrated GPU is defined as a GPU located on a monolithic silicon die along with a CPU and memory controller, behind a unified memory subsystem. Performance tests are conducted using specific computer systems and reflect the approximate performance of MacBook Pro."
The definition basically excludes Ryzen APU's since the Ryzen chiplet is on a separate silicon die from the CPU and memory controller
Ahaha, sneaky Apple marketing
Yeah, we desperately need some benchmarks. It's probably a 15W Tiger Lake, since I don't think 30W parts are commercials available?I'm not even sure about that. It said the product was tested in October 2020 with commercially available at the time of testing. I don't believe the 96 EU Tiger Lake part was available
I believe you are correct. They weren't available yet - heck, Tiger Lake 15W was barely shipping in October 2020 and it said "at the time of testing" which could have been October 1st!Ahaha, sneaky Apple marketing
Based on other indirect evidence (flops, comparisons to previous models), I'd expect the highest-performing M1 (estimated 15W) to be a bit faster than the 30W Tiger Lake Xe in synthetics, and 30-50% faster in real-world gaming.
I think important thing to consider is that this marketing stuff is not written for us nerds. It's written for an average consumer. It is entirely possible that the real performance will disappoint, but based on what we saw so far I choose to be optimistic.
Yes, I would assume the comparison is against the same Ice Lake 28w chip in the Intel Macbook Pro, which I think in fact was the fastest shipping integrated chip in early October (at least based on the careful wording they used). That meets expectations and is in line with what leman said. Tiger Lake GPU is 2xish faster than Ice Lake, and that would make the M1 a bit faster than that.
I wouldn't expect it. Right now RT is only available on mid-to-high end GPUs (of which have ridiculous TDP and cost nearly as much or more than these new Macs). Expecting Apple to come out with a laptop SoC with competitive RT is unrealistic.
Yep. As we said before, the A12Z is already faster than the 28W Intel Xe at 3DMark Wild Life. The M1 should beat any competing laptop iGPU with ease.
Interesting observation: Apple does claim RT support in Metal for M1, but I doubt that it has dedicated RT support hardware. Some limited RT for basic shadows etc. would still be possible in games, would be interesting to run some benchmarks. This is a similar approach Chrysis Remastered uses, only that Metal has some advanced shader support to make it happen (like recursive shader functions etc.)
Yeah I saw that in the presentation. Sadly none of the 3dmark benchmarks that support RT work on the Mac or iOS so 🤷🏽♂️. I doubt GFXBench will ever add RT support but would love to be surprised.
SotTR has RT support on the PC, maybe Feral can release a patch to add it to the Mac.
It's not. I mean, A14 might be slightly more efficient with recursive shader invocation and async compute that one needs for this functionality, but as I mentioned before, I won't expect any fixed-function intersection hardware.
It's important to stress however that DX12 and Nvidia raytracing is also implemented in compute shaders — they just have some dedicated hardware to find ray-geometry intersection faster.
Tech Talks - Videos - Apple Developer
At 2:40
OH, ok, I thought I missed something yesterdayTech Talks - Videos - Apple Developer
developer.apple.com
At 2:40
It's not. I mean, A14 might be slightly more efficient with recursive shader invocation and async compute that one needs for this functionality, but as I mentioned before, I won't expect any fixed-function intersection hardware.
It's important to stress however that DX12 and Nvidia raytracing is also implemented in compute shaders — they just have some dedicated hardware to find ray-geometry intersection faster.
Tech Talks - Videos - Apple Developer
At 2:40
New 3DMark test measures pure ray-tracing performance
Find out more at benchmarks.ul.com
benchmarks.ul.com
of course none of these are currently available on macOS so I dunno how one can test the RT abilities.
is supposed to isolate RT performance from regular rasterization performance. I think Quake RTX and Minecraft RTX are also able to approximate RT performance while minimizing rasterization performance differences.New 3DMark test measures pure ray-tracing performance
Find out more at benchmarks.ul.com
of course none of these are currently available on macOS so I dunno how one can test the RT abilities.
You can download the free source code of one of WWDC RT demos and compile it in Xcode for example
I wouldn't expect UL to go through the trouble of porting these benchmarks to Apple Silicon, because... why would they? Nobody really cares except