• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Question Intel Mont thread

Page 9 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
I think most impressive part of SRF is "Intel 3". It seems the tweaked EUV process is showing potential. 18A should be exciting as that will be tweaked from 20A "test" node. chadmont plus server chip on it should be more than competitive against Turin-D.
 
We missed one interesting slide from the presentation with a comparison between Conroemont and Raptor Cove on different power levels...

That looks insane!
 

Attachments

  • test.png
    test.png
    711.7 KB · Views: 37
We missed one interesting slide from the presentation with a comparison between Conroemont and Raptor Cove on different power levels...

That looks insane!

With Skymont being roughly equal in IPC but on a much more efficient node, this isn't surprising. Raptor Cove would eventually take over in performance if the graph extended further, but it would obviously take a whole lot more power to get there.
 
Still, it's quite the leap from Gracemont/Crestmont. Gracemont ran too hot. Crestmont was clobbered by tile latency. I wonder what Skymont's Achilles heel will be.

For sure, the leap they made with Skymont is extremely impressive. I'm very interested in seeing the final clocks as well as we still don't have official confirmation on that, unless I missed it. For its purposes, I don't know if Skymont will have an Achilles' heel. Against Zen 5 it will have some "weaknesses" (e.g., lacking AVX-512) but I don't think that they will matter for its intended customers.
 
Last edited:
Imagine Alder Lake-N successor, but with Skymont...
Yeah but unfortunately, Intel is too dumb and too greedy to do something like that. If a $300 laptop can give you that level of performance, why would anyone bother buying their Core Ultra? Sadly they will continue spamming the market with anemic Gracemont 4-core laptops and letting OEMs charge up to $500 for those netbooks.
 
Yeah but unfortunately, Intel is too dumb and too greedy to do something like that. If a $300 laptop can give you that level of performance, why would anyone bother buying their Core Ultra? Sadly they will continue spamming the market with anemic Gracemont 4-core laptops and letting OEMs charge up to $500 for those netbooks.
Issue with these "Mont" is gpu support, without proper gpu support nothing will change.
 
Alder Lake-N has the same IGP, just less EUs.
It needs ARC iGPU with ability to at least play some popular esports titles at 30fps minimum (no frames dropped). They can strip out the RT and XMX units. But basic raster support at decent fps with updated drivers to at least display something on the screen (even if at 5 fps) in the latest games is a must. It can be a way to sell cheap external gpu docks to netbook owners (like a lot of underprivileged kids). Even an ARC A380 in a dock would be a huge improvement for such a kid.
 
No it doesn't. Not every product is meant for gaming.
I'm of the opinion that people shouldn't be punished for not having enough money. A decent iGPU can come in handy any time. Grandpa/grandma or the little kid could want to explore some beautiful game. They should not be disappointed with a game refusing to launch or running like molasses even at the lowest settings. I think this is something that Snapdragon may excel at. Their value GPUs may be faster than whatever Intel pairs with the N CPUs.
 
I'm of the opinion that people shouldn't be punished for not having enough money.
Then don't entertain financial punishment on poor people who want an affordable compute device. These chips are made with cost in mind, specifically to enable all kinds of compute devices while adding as little to BoM as possible. Once you introduce gaming as a performance threshold for these chips, their GPU becomes the defining feature and takes up most of the die area along the uncore. The choice of CPU cores becomes almost irrelevant.
 
Then don't entertain financial punishment on poor people who want an affordable compute device. These chips are made with cost in mind, specifically to enable all kinds of compute devices while adding as little to BoM as possible. Once you introduce gaming as a performance threshold for these chips, their GPU becomes the defining feature and takes up most of the die area along the uncore. The choice of CPU cores becomes almost irrelevant.
But they could provide pcie lanes for us to use our own lowend gpu, even something like GT 1030 would make these cpu much better.
 
But they could provide pcie lanes for us to use our own lowend gpu, even something like GT 1030 would make these cpu much better.
Yeah it's Intel that's the problem here and hopefully Snapdragon or Mediatek will create some competition in this space. Even if Intel cannot include a stronger iGPU to keep the core size small, they should at least let the laptop have some basic upgrade functionality. Four lanes of PCIe 4.0 could be enough bandwidth for a low end card that gives options to netbook owners and lets them keep the device longer instead of chucking it in the trash when it can no longer grow with their needs.
 
Even the top of the line Kompanio 1380 has a pretty mediocre GPU.
Hey, maybe their "secret" deal with Nvidia is to bring the Geforce 1030 to their SoC 😛

There's also the possibility of a slightly weaker GPU than the Switch, since Nintendo will be moving onto a better GPU with Switch 2 so makes sense for Nvidia to milk their old IP for some more moolah.
 
There's also the possibility of a slightly weaker GPU than the Switch
The Switch? 😅

Switch's TX1 uses a 2015 ultra low end Maxwell GPU closer to the barebones 1 WGP/2 CU RDNA2 offering in Raphael and Granite Ridge than anything remotely mid/hi end in modern ARM based SoCs, while being far behind in feature set.

Going by compute throughput alone the SDXE gen 1 Adreno GPU is about 8 times faster at over 4 TFLOPS.
 
The Switch? 😅

Switch's TX1 uses a 2015 ultra low end Maxwell GPU closer to the barebones 1 WGP/2 CU RDNA2 offering in Raphael and Granite Ridge than anything remotely mid/hi end in modern ARM based SoCs, while being far behind in feature set.

Going by compute throughput alone the SDXE gen 1 Adreno GPU is about 8 times faster at over 4 TFLOPS.

Yeah, there is not a snowball's chance in hell that a Mediatek PC chip uses something weaker than a downclocked X1.
 
Back
Top