Intel Meteor Lake Arc Graphics: A Fantastic Upgrade, Battles AMD RDNA3 Integrated Graphics

[darkbasic]

Finally some competition... So bad that the CPU side sucks.

[blackshard]

If I understood correctly, the AMD Framework uses DDR5-5600 memory and the Intel uses some variant of LPDDR5x (I have no idea which - do you know Michael?). Would love to see a comparison when AMD APU is connected to LPDDR-6400 as that would alleviate at least some of the bandwidth discrepancy. Still, Intel seems to have invested more die space for iGPU than AMD did in Phoenix.

That's an interesting point, memory bandwidth is essential in graphics. Bandwidth may be affected by single channel vs. dual channel issue also, nontheless Intel graphics is still performing plenty well if we consider the overall efficiency. This and the very good support for hardware video encoders are the main selling points of this processor family.

[AdrianBc]

Read the graphs carefully: Intel beats AMD in gfx while usually drawing less power.

While, as expected, the new Intel GPU, with 4/3 more execution units and a clock frequency that is higher than 3/4 of the AMD clock frequency beats the AMD GPU, while consuming less, we should not forget that the Intel GPU is made by TSMC, unlike the Intel CPU, which is made by Intel.

Therefore the good performance of the Intel GPU only demonstrates the quality of the Intel circuit designers, not of the Intel manufacturing capability.

We do not know which are the differences between the TSMC process used for the Intel GPU and the process used for AMD Phoenix.

Because Meteor Lake has just been launched, it might use an improved TSMC process, which may be the cause of its better power consumption.

The 2024 variant of mobile Zen 4, the 8040 series, is said to consume slightly less than the 2023 7040 series, presumably also due to using a more recent TSMC process variant.

[mikk]

Who would expect that top CPU with maximum turbo power of 115W (or assured power of 65W) beats CPU for ultrabooks by astonishing 8%!

You should be really crazy to buy anything from team blue today. Unless you have free electricity and need some high-tech heater in your house LMAO.

Theoretical maximum power numbers are meanignless for laptops, you always need to check the real power numbers. Maybe read the test properly.

For the initial integrated graphics benchmarks run, here is a look at the SoC power consumption over the entire span of tests conducted. The Core Ultra 7 155H on average was consuming 24 Watts to the Ryzen 7 7840U at a 25.8 Watt average. The peak consumption was also lower for Meteor Lake on the Acer Swift Go 14 with a 43.5 Watt peak compared to 51 Watts on the Framework 13. Not only was the Core Ultra 7 155H pulling slightly less power but it was a big upgrade over the Alder Lake MSI EVO laptop with the Core i7 1280P having a 36 Watt average and 67 Watt peak.

That's an interesting point, memory bandwidth is essential in graphics. Bandwidth may be affected by single channel vs. dual channel issue also, nontheless Intel graphics is still performing plenty well if we consider the overall efficiency. This and the very good support for hardware video encoders are the main selling points of this processor family.

In the tests I have seen there is not much difference between 5600 and 7467 on Meteor Lake, don't expect more than 5-10%.

[pete910]

To be fair that's a damn good showing from Intel !

[CombatEmu]

If Battlemage does come out on schedule next year, I'm very curious to see how well it performs. The CPU side of Intel is a joke (and those marketing slides were even more so), but at least the GPU side is actually making progress. Allegedly, RDNA4 will only target midrange and if Battlemage is competitive, then it may drive prices down a bit in that segment.

[pWe00Iri3e7Z9lHOX2Qx]

It wasn't very long ago that AMD had significantly worse cores and significantly better iGPUs than Intel. The role reversal is interesting.

[fahrenheit]

If I understood correctly, the AMD Framework uses DDR5-5600 memory and the Intel uses some variant of LPDDR5x (I have no idea which - do you know Michael?). Would love to see a comparison when AMD APU is connected to LPDDR-6400 as that would alleviate at least some of the bandwidth discrepancy. Still, Intel seems to have invested more die space for iGPU than AMD did in Phoenix.

That is correct, APUs run faster with LPDDR5 memory, just check the Rog Ally review here: https://www.phoronix.com/review/rog-...indows-linux/2 and compare the Gravity Mark vulkan renderer results.
Probably a more apples to apples test would be testing a 7940H(S) running LPDDR5 memory vs this. Because the Core Ultra U range, which are the natural competitors to the Ryzen U range, have at most 4 XE execution units.

On the other hand on the intel side, gen over gen improvements taking into consideration the H range of cpus is great.

[SomeoneElse]

Notebookcheck did also some tests und luckily there some bandwidth numbers for the (LP)DDR5 RAM:
- Acer-Swift-Go-14: https://www.notebookcheck.net/Acer-S....783349.0.html
- Framework Laptop: https://www.notebookcheck.net/Framew....756613.0.html
- HP-EliteBook-840 with 1280P: https://www.notebookcheck.net/HP-Eli....674583.0.html

----

Laptop / CPU	Acer / 155H	Framework / 7840U	HP / 1280P
RAM Type	LPDDR5-6600	DDR5-5600	DDR5-4800
Copy Bandwidth (AIDA64 / Notebookcheck)	80515 MB/s	62500 MB/s	56337 MB/s
Relative Bandwidth	100%	~77.6%	~70%

---
Edit, the RAM type/speed of the Acer could be wrong, imho is an indication that it could be LPDDR5-6600 based on the CPU-Z Screenshot from Notebookcheck, see: https://www.notebookcheck.net/filead...-14-120008.png

[RejectModernity]

Wake me up when Intel will be capable of running real modern games.
Former A770 pain-in-the-ass enjoyer