Qualcomm revealed X Elite's benchmark scores

[Xiao_Xi]

There is a GB6 score of the 12-core Qualcomm SoC: 2500 in single-core and 12500 in multi-core.

Qualcomm ZH-WXX - Geekbench

Benchmark results for a Qualcomm ZH-WXX with a Snapdragon X Elite - X1E80100 - Qualcomm Oryon processor.

browser.geekbench.com

 

[thenewperson]

There is a GB6 score of the 12-core Qualcomm SoC: 2500 in single-core and 12500 in multi-core.

Qualcomm ZH-WXX - Geekbench

Benchmark results for a Qualcomm ZH-WXX with a Snapdragon X Elite - X1E80100 - Qualcomm Oryon processor.

browser.geekbench.com

M2 Pro level 🤔

 

[MrGunny94]

M2 Pro level 🤔

Yeah but what I'm not expecting is the same type of battery life.. I'd like to be wrong though..

 

[leman]

There is a GB6 score of the 12-core Qualcomm SoC: 2500 in single-core and 12500 in multi-core.

Qualcomm ZH-WXX - Geekbench

Benchmark results for a Qualcomm ZH-WXX with a Snapdragon X Elite - X1E80100 - Qualcomm Oryon processor.

browser.geekbench.com

I hope this is not representative of the final products performance. That would be extremely disappointing.

 

[Homy]

There are 5 results, the lowest with 11010 in multi-core. The singel-core result is like 8-core M2 MBA and the multi-core is like 8-core M3 iMac.

Geekbench Search - Geekbench

browser.geekbench.com

 

[dgdosen]

There is a GB6 score of the 12-core Qualcomm SoC: 2500 in single-core and 12500 in multi-core.

Qualcomm ZH-WXX - Geekbench

Benchmark results for a Qualcomm ZH-WXX with a Snapdragon X Elite - X1E80100 - Qualcomm Oryon processor.

browser.geekbench.com

Ship it!

That's pretty close to an Apple Silicon experience. I hope thermals/battery life come close as well. I also hope they support a Linux variant when they ship.

 

[Sydde]

Of course it is going to be faster than M3 in multi-core. It's a 12P-core design vs. 4P+4E design. Single-core though? Unlikely. They will be close though (assuming Oryon can actually maintain its boost speed).

This GB page for the Qualcomm CRD model disagrees with you. It says the processor has one core, with 12-thread SMT. That would be 50% more SMT than IBM has on its Power10 cores. Assuming, of course, that GB6 was not somehow tricked into only seeing a single core.

 

[deconstruct60]

This GB page for the Qualcomm CRD model disagrees with you. It says the processor has one core, with 12-thread SMT. That would be 50% more SMT than IBM has on its Power10 cores. Assuming, of course, that GB6 was not somehow tricked into only seeing a single core.

The motherboard ID is not supported.
Some Linux Os version that isn’t released
( https://wiki.debian.org/DebianTrixie. )

Cluster 1 has zero cores .

It isn’t hard to pump fake data to Geekbench. 12 way SMT should more so be a give away that this is hokum , not some announcement of threading prowess over Power 10 cores. POWER 10 cores actually exist, while this entry is likely from an alternative universe .

 

[leman]

This GB page for the Qualcomm CRD model disagrees with you. It says the processor has one core, with 12-thread SMT. That would be 50% more SMT than IBM has on its Power10 cores. Assuming, of course, that GB6 was not somehow tricked into only seeing a single core.

That’s not what Qualcomm has announced last year. I think it makes more sense to trust their representative than whatever GB reports at this point (especially since this is experimental, less supported hardware).

 

[MrGunny94]

I hope this is not representative of the final products performance. That would be extremely disappointing.

I don't think it's that bad especially for a first gen product, plus it could improve with additional improvements over drivers and etc.. We are still far away from Q3 2024

 

[leman]

I don't think it's that bad especially for a first gen product, plus it could improve with additional improvements over drivers and etc.. We are still far away from Q3 2024

Yeah, it's lower than the numbers measured at the event. Could be an unoptimized version of Windows...

 

[Xiao_Xi]

I also hope they support a Linux variant when they ship.

Linaro and Qualcomm are working on it.

Qualcomm and Linaro Enable Latest Flagship Snapdragon Compute SoC | Blog | Linaro

Running fully fledged Debian desktop (including GPU rendering) with patches on top of linux-next on the X Elite (X1E80100) platform even before it has been announced

www.linaro.org

x1e80100 - search results

lore.kernel.org

 

[Closingracer]

Is not a first product..they just changed their name for a more dramatic "looks". But im glad after 3 some kind failures they keep pushing the world of Windows to arm
View attachment 2301582

I hada Dell laptop with the Gen 2 and it was ok.... its problem was more of Dell putting horrible specs in it vs the SOC. Heck that was the best battery life I have seen on a Windows laptop if I'm being honest. To bad it was hampered but low memory and meh screen

 

[ricom2ger]

A new GB leak.

 

[jdb8167]

A new GB leak.

View attachment 2354505

If that single-core result is correct, that's pathetic. My M2 gets over 2600 SC.

 

[Homy]

A new GB leak.

View attachment 2354505

M1 has 18758 in OpenCL:

 

[MrGunny94]

A new GB leak.

View attachment 2354505

Curious to see this running under Linux...

 

[Xiao_Xi]

A new GB leak.

This result seems fishy. The SoC (Snapdragon X Elite - X1E78100) of this notebook is different from the SoC (Snapdragon X Elite - X1E80100) of the notebook of the previous GB scores.
browser.geekbench.com/v6/cpu/5076297?baseline=5019569

There is no mention of this model in the Linux kernel,

https://lore.kernel.org/all/?q=X1E78100

But there are hundreds for the above model.

X1E80100 - search results

lore.kernel.org

 

[ricom2ger]

M1 has 18758 in OpenCL:

View attachment 2354550

Buggy Andreno drivers kick in. My Acer Aspire 1 with Qualcomm CPU couldn't even finish this benchmark.

 

[MiniApple]

according to the newest Windows Central podcast and Zac Bowden has been hearing that:

• all (so Intel and Qualcom ARM) Surface Pro 10 and Laptop 6 models could be announced at the March 21, 2024 event.
• the release will still be staggered, with the Qualcom ARM X Elite options being available from June on wards due to the necessary (underlining) new Windows on ARM version only being available then.

 

[Chuckeee]

according to the newest Windows Central podcast and Zac Bowden has been hearing that:

• all (so Intel and Qualcom ARM) Surface Pro 10 and Laptop 6 models could be announced at the March 21, 2024 event.
• the release will still be staggered, with the Qualcom ARM X Elite options being available from June on wards due to the necessary (underlining) new Windows on ARM version only being available then.

I wonder if other manufacturers (Dell?, Lenovo?) are planning to come out Qualcomm Elite ARM laptops? I would be particularly interested in a small Linux Laptop. But from years of previous experience I just could not stomach buying another Microsoft product, especially if there’s a way I could avoid it. Just too much bad blood for me to forgive and forget

 

[rampancy]

according to the newest Windows Central podcast and Zac Bowden has been hearing that:

• all (so Intel and Qualcom ARM) Surface Pro 10 and Laptop 6 models could be announced at the March 21, 2024 event.
• the release will still be staggered, with the Qualcom ARM X Elite options being available from June on wards due to the necessary (underlining) new Windows on ARM version only being available then.

I admit I haven't been keeping up but how are they going to handle running non-native software? Will Windows on ARM come with a binary translation layer like Rosetta 2?

Someone also mentioned bad display drivers being an issue, too. I wonder how much of an impact that might have on performance in the end.

 

[leman]

I admit I haven't been keeping up but how are they going to handle running non-native software? Will Windows on ARM come with a binary translation layer like Rosetta 2?

Windows on ARM has offered this functionality from the start. There is a catch though. Apple Silicon can emulate certain tricky aspects of x86 behavior in hardware, this is why Rosetta 2 is both fast and reliable. Microsoft does not have this. Their translation layer can be as fast, but it cannot ensure correct execution of some multithreaded applications. So Microsoft has implemented multiple compatibility modes that ensure correctness, but come with significant cost to performance. What this means is that simple software will work, some complex software might crash or not operate correctly, and it’s up to the user to select the suitable correctness/performance tradeoff. Not the best user experience.

 

[MiniApple]

I wonder if other manufacturers (Dell?, Lenovo?) are planning to come out Qualcomm Elite ARM laptops? I would be particularly interested in a small Linux Laptop. But from years of previous experience I just could not stomach buying another Microsoft product, especially if there’s a way I could avoid it. Just too much bad blood for me to forgive and forget

I believe I've read somewhere that Qualcom have said that it will be Windows PCs only at launch, but they are open to other OS, like Linux.

Linux Community surely will find a way too eventually, provided the Qualcom performance/battery life is there and makes it desirable (see Asahi Linux on the Apple M's)

 

[jdb8167]

Windows on ARM has offered this functionality from the start. There is a catch though. Apple Silicon can emulate certain tricky aspects of x86 behavior in hardware, this is why Rosetta 2 is both fast and reliable. Microsoft does not have this. Their translation layer can be as fast, but it cannot ensure correct execution of some multithreaded applications. So Microsoft has implemented multiple compatibility modes that ensure correctness, but come with significant cost to performance. What this means is that simple software will work, some complex software might crash or not operate correctly, and it’s up to the user to select the suitable correctness/performance tradeoff. Not the best user experience.

It's amazing to me that the CPU architects that came from Apple and are designing a brand new processor from scratch didn't bother implementing something like Apple's TSO mode. Are we sure they didn't? If not it calls into question how expert these ex-Apple engineers really are. To not have reliable x86 and x86-64 emulation without heavy use of memory fences that slows down execution is borderline incompetent.