Both Intel and AMD will catch up Qualcomm in efficiency before Qualcomm catches them at performance, just watch.
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.
Qualcomm revealed X Elite's benchmark scores
- Thread starter sunny5
- Start date
- Sort by reaction score
In what sense for either? If we're talking native ST/MT performance, then Qualcomm is already competitive with AMD and the currently available Intel processors if not well ahead of the latter's Meteor Lake. Intel is claiming to retake the lead in ST performance and MT perf-per-watt (though not MT performance) with their soon-to-released Lunar Lake processors - though as already noted 1st party benchmarks on unreleased processors should never be assumed to be accurate - and AMD already has roughly equivalent MT perf/W to the Elite (depending on workload). Neither of them are anywhere close to Qualcomm's ST perf/W and that's unlikely to change anytime soon. So I'm not quite sure where either statement is coming from? Emulated performance?
Last edited:
What do you mean? Qualcomm is already as fast or faster than top x86 laptop chips.
Well, Intel are making some impressive claims for their latest "Lunar Lake" low-power processors. Of course, Elite X is on N4 while Lunar Lake is on N3.
It seems that Intel's current poor financial situation may open up new opportunities for AMD and Qualcomm.
It is curious how Intel-only notebook manufacturers are now willing to use CPUs from manufacturers other than Intel.
It is curious how Intel-only notebook manufacturers are now willing to use CPUs from manufacturers other than Intel.
it's ironic, now that Intel FINALLY seem to have products (Lunar Lake and beyond aka Core Ultra 200 and later) that not only deliver very good performance, but also very good efficiency and battery life while offering good graphics too...
Last edited:
The main obstacle for Qualcomm is the state of Arm support on Windows.
When something like the Adobe Creative suite only supports a few of the main applications and only the older 2023 version to boot, the road will be long as it has been with Apple in this transition period.
32-bit is also going to be a major problem going forward, so hopefully the proposed x86S ISA will take hold and applications will get updated to 64-bit.
When something like the Adobe Creative suite only supports a few of the main applications and only the older 2023 version to boot, the road will be long as it has been with Apple in this transition period.
32-bit is also going to be a major problem going forward, so hopefully the proposed x86S ISA will take hold and applications will get updated to 64-bit.
Their top iGPU Arc 140V in Ultra 9 288V is only as fast as the base M3 8-core GPU.
How do you know? I've only seen the gaming performance comparison published by Intel and Apple's GPU wasn't mentioned there.
...which would make them tied for fastest "normal" iGPU ever, until the Mac M4 ships. Pretty decent.
By normal, I mean, "not Mx Pro or Mx Max", as neither of those chips is a reasonable choice for typical consumers.
The GPU is the same size across the current (mobile) lineup, other than some low-end (binned, I guess) models that have 7 cores instead of 8. Remember that the GPU is on a separate die from the CPU, and it is fabbed on TSMC N6, whereas the Apple GPU is on the same die as everything else and so fabbed at N3.
You're thinking of Meteor Lake. For Lunar Lake, the GPU is on the compute tile with the CPU and NPU and therefore manufactured on N3B. Only the SOC tile (controllers and IO) is manufactured on N6 and the base tile that glues everything together is Intel.
Lunar Lake - Wikipedia
Last edited:
Actually base M3 in MBA comes with 10-core GPU too but all Lunar Lake Ultra 9/7 variants have 8 cores. The cheaper models come with downclocked 140V so M3 could already be faster and remember Intel said ”World’s fastest built-in GPU” which is only true if you don’t compare with Apple. Funny that they suddenly don’t mention Apple.
Last edited:
”Gelsinger also confirmed the expansion of orders to TSMC, confirming that TSMC will hold orders for Intel's Arrow and Lunar Lake CPU, GPU, and NPU chips this year, and will produce them using the N3B process, officially ushering in the Intel notebook platform that the outside world has been waiting for many years. Past leaks have indicated that Intel's Arrow Lake processor family will have CPU tiles based on their in-house 20A process, while TSMC takes care of the GPU tile aspect with their 3 nm N3 process node.”
Last edited:
Intel compared the iGPUs using their performance in various games. I doubt the M3 GPU would fare well in that comparison.
I doubt your doubt, assuming you could get the games running at all, even in emulation. But I would love to see some numbers.
Do you know how awesome it is to finally have even M-like performance (with little to no noise) WITH good efficiency and battery life in the Windows and Linux world?!
This did not exist prior to Qualcom X Plus/Elite, AMD 300 and Intel 200.
the iGPUs are finally good enough, that I can sell my small xbox console and play my game library on these devices too with similar settings.
Unfortunately NBC didn't put the gaming M3 results in the article itself, except for efficiency and benchmarks, but digging them up from the M3 10-core article we can compare (I'm not going to do all of them, too many but you can do that yourself - also this is the 15" Air not the 14" pro so the full M3 will be somewhat constrained in both energy usage and performance with no active cooling in some of these benchmarks):
AMD Zen 5 Strix Point iGPU analysis - Radeon 890M versus Intel Arc Graphics, Apple M3 and Qualcomm Adreno X1-85
Notebookcheck analysis of the new AMD Radeon 890M iGPU compared with the Intel Arc Graphics, Apple M3 and the Qualcomm Adreno X1-85.
Apple MacBook Air 15 M3 review - Apple’s large everyday MacBook gets a power up
Notebookcheck reviews the new Apple MacBook Air 15 with the M3 SoC, 16 GB of RAM and Wi-Fi 6E.
The Witcher 3 (non-native for M3) 1080p Ultra Hairworks On:
| Frame Rate | Energy | |
| 890M | 31-34 FPS | 45-76W |
| M3 10-core (Air 15") | 24 FPS | 21W |
Baldur's Gate (native for M3) 1080p Ultra:
| Frame Rate | |
| 890M | 24.9-29.6 FPS |
| M3 10-core (Air 15") | 19 FPS |
The ranges on the 890M is to denote the different models with different performance/power profiles, but overall we can see that the 890M is a bigger, more powerful, more power hungry GPU. At the same time, it's true that the Apple M3 matches it or beats it in several benchmarks though not in the games above (more below). Also note that NBC had slightly lower scores than Just Josh for their 8C M3 Extreme Wildlife test, but sometimes these tests have high variance, especially in an Air due to the more thermally constrained environment. They also had higher Wildlife scores for the P16 than the P13 (though those two were closer elsewhere). Unfortunately NBC did not have Metal scores for GB compute, but honestly I've found those scores to be difficult to compare across APIs despite Primate Lab's insistence that it can be and with OpenCl being deprecated on AS ... well ... I dunno I still refer to it, but I usually have a small asterisk next to the results.
Now to Intel and their claims and the slides here:
Slide 31 for the GPU comparison with the HX 370, Slide 87 for the GPU Notes.
Look closely at the Slide 31 graph, the games are all run at 1080p medium, basically they are testing these games in the least likely to be GPU limited scenario even for the small iGPUs tested.
For instance, according to NBC, here is the performance of the M3 and S16 on Total War Pharaoh 1080p for multiple presets:
| Ultra Frame Rate | High Frame Rate | Medium Frame Rate | |
| 890M (S16) | 42.2 FPS | 50.7 FPS | 67.9 FPS |
| M3 10-core (Air 15") | 35 FPS | 44 FPS | 83 FPS |
Once the graphics fidelity is reduced far enough, the CPU performance enters the mix and begins to dominate, and the M3 15" Air overtakes the S16. Further Intel picked S16 which is often, though not always, the lowest performing HX 370 device. Not every game shows this behavior at Medium mind you, Baldur's Gate does not even at Low!, and Total War is a particularly CPU intensive title, but it is far more likely to do so.
Intel claims to have tested the M3 in making its "Best iGPU" statement, but obviously don't show the results anywhere nor do they show the non-gamin graphics benchmark results (merely stating they were "exceptional"). Also they claim the S16 gets 83 FPS in Total War 1080p Medium (as opposed to the 67.9 above) which is more similar to the P13/16 and may reflect differences in how the benchmark was run in terms of performance profile though it didn't seem that way. Unclear what's going on.
One more thing, Intel claims on slide 33 and 34 that the matrix accelerators help with upscaling and maybe ray tracing, similar to Nvidia's RT/DLSS, and is one reason why I had hoped that even for the base M4 that Apple would've added matrix accelerators to the GPU. Oh well.
As always, wait for 3rd party reviews.
Last edited:
According to Intel's slide, the Intel SoC achieves 59 FPS and the AMD SoC, 46 FPS at 1920x1080 at medium settings. PCMag reports that the 14-inch Apple MacBook Pro (2023, M3 Pro) achieves 57 FPS at 1920x1200 at medium settings.
Testing Apple M3 Graphics: Gaming on the Mac, From M3 to M3 Max
With four of the latest Macs on the test bench, we conducted an early investigation into how the latest M3 Apple Silicon helps make Macs more viable for gamers.
It is possible that the iGPU in Intel's new SoCs will perform more like the M3 Pro GPU than the M3 GPU in gaming.
NBC has the FPS at 44 FPS for the base M3 at 1080p Medium (with of course the caveat the Shadow is Rosetta which might matter at 1080p Medium and it's the 15" Air).According to Intel's slide, the Intel SoC achieves 59 FPS and the AMD SoC, 46 FPS at 1920x1080 at medium settings. PCMag reports that the 14-inch Apple MacBook Pro (2023, M3 Pro) achieves 57 FPS at 1920x1200 at medium settings.
Testing Apple M3 Graphics: Gaming on the Mac, From M3 to M3 Max
With four of the latest Macs on the test bench, we conducted an early investigation into how the latest M3 Apple Silicon helps make Macs more viable for gamers.www.pcmag.com
It is possible that the iGPU in Intel's new SoCs will perform more like the M3 Pro GPU than the M3 GPU in gaming.
Apple MacBook Air 15 M3 review - Apple’s large everyday MacBook gets a power up
Notebookcheck reviews the new Apple MacBook Air 15 with the M3 SoC, 16 GB of RAM and Wi-Fi 6E.
They have the M3 Pro at 66 FPS at 1080p Medium (better thermals, still Rosetta).
Apple MacBook Pro 14 2023 M3 Pro review - Improved runtimes and better performance
Notebookcheck has tested the MacBook Pro 14 with the 11-core M3 Pro, 18 GB RAM and a 512-GB SSD which costs 2,499 Euro.
Thus I'd say it's in between, closer to a binned M3 Pro. For this title anyway. From slide 80 we can see that this variant of the Lunar Lake is the 288V which is Intel's "30W" version of the SOC Turbo 37W (basically drawing the same power as the M3 Pro, almost certainly more given Intel's history of TDP ratings). For comparison, the (EDIT)
Compared the base M3 10-core the 288V has two fewer cores but 46% faster clocks. Assuming all else being equal, it won't be so this shouldn't be taken as anything other than a vague generality, that's an expected 17% better performance but for the cost of a sizable power increase as its speed comes from clocks not cores. I'm unclear about the architectural details of Intel's GPU - I suspect it's closer to Nvidia/AMD than Apple which may give Apple a decent boost in relative raster performance depending on the title but obviously more games are likely to be optimized for Windows/DX than macOS/Metal if they even have a macOS/Metal port (and sometimes even if they do, the non-native port is actually better). So that advantage may cancel out
Last edited:
That was my point. Before they always used to compare their chips’ performance and power efficiency and battery life with M1/M2 but now not only no mention of Apple but also only gaming benchmarks so perhaps M3 results are not in favor of Lunar Lake. In fact Dave2D shared some official Intel benchmarks and M3 8c has higher single score (142) in Cinebench 2024 than the top Ultra 9 288V (130).
Games has always run faster in Windows than macOS but iMac M3 10-core GPU can do 45-80 fps at 1080p Medium in Metro Exodus. It has two more GPU cores but Metro Exodus is also a x86 game on Apple Silicon so it’s not running natively. Don’t know how they benchmarked the game with Ultra 9 288V but it does 45 fps.
M3 10-core GPU runs Total War: Pharaoh siege benchmark at 83 fps 1080p Medium, about the same as AMD 890M. Again not sure if Intel has used the same benchmark but Ultra 9 does 102 fps.
According to Intel's slide, the Intel SoC achieves 59 FPS and the AMD SoC, 46 FPS at 1920x1080 at medium settings. PCMag reports that the 14-inch Apple MacBook Pro (2023, M3 Pro) achieves 57 FPS at 1920x1200 at medium settings.
Testing Apple M3 Graphics: Gaming on the Mac, From M3 to M3 Max
With four of the latest Macs on the test bench, we conducted an early investigation into how the latest M3 Apple Silicon helps make Macs more viable for gamers.
It is possible that the iGPU in Intel's new SoCs will perform more like the M3 Pro GPU than the M3 GPU in gaming.
Total War: Three Kingdoms is a x86 game running through Rosetta. M3 Pro 14c does 125 fps in the Pharaoh siege benchmark while Ultra 9 does 102 fps.
Last edited:
AMD 890M does 81.9 fps at 1080p Medium so it's on par with M3 10-core GPU in Pharaoh.
Yes, I can imagine but wait until real benchmarks come out before switching.
Yes according to NBC the P13/16 variant of HX370 scored that, but the S16 variant, the one tested by Intel and why I restricted the final chart to only that number, only got 67.9FPS though Intel claimed their S16 did 83. NBC's S16 had lower performance in general than the HX370 in the P13/16 (with a few exceptions where it matched or beat it). Regardless the main point was to show testing at 1080p medium can skew the results to allow a weaker GPU to catch up to a bigger one.
This shows up in your chart as well comparing low, medium, high, and ultra performance the M3 10-core does much better the lower the graphics setting as the strength of Apple's CPU begins to outmuscle AMD's. So it is entirely possible that Intel wins at medium but gets progressively worse at high and ultra relative to the AMD chip. Also from NBC's analysis, the previous Intel GPU, which may have different behavior, scored much worse than linear scaling at lower TDP. Since Intel only showed results from its highest GPU configuration, it is unclear how a more comparable GPU to the M3 in terms of power would perform - i.e. the Ultra 5 226/228V.