Intel N100 Quad E-Core Gaming Performance Assessed

[T0@st]

Team Pandory has tested the gaming potential of an Intel Alder-Lake-N SoC—not many outlets have bothered to give the N100 much coverage in this aspect, since the chip's makeup is E-core only and it only offers single-channel memory support. Team Blue has emphasized power efficiency rather than raw performance with its super low budget 2022 successor to old Pentium and Celeron processor product lines. The utilization of modern Gracemount CPU cores does it some favors—notably granting L3 cache support, but the chip has been designed with entry-level productivity in mind.

Naturally, in-game testing focuses attention on the N100's integrated GPU, based on Team Blue's Xe-LP architecture—it features 24 execution units (EUs), support for AV1 decode capabilities, and 8K 60 FPS video playback. Arc Alchemist offers roughly double the performance when compared to the Xe-LP iGPU, so we are not expecting a big "wow factor" to be delivered by the plucky Alder-Lake-N SoC (6 W TDP). Team Pandory benchmarked a laptop sporting a single stick of 8 GB DDR5 RAM and the N100 quad E-core CPU (capable of 3.4 GHz turbo boosting), with 6 MB of L3 cache. The ultra portable device was able to hit 60 FPS in a couple of older games, but the majority of tested titles ran at 20 to 30 20 FPS (on average). Graphics settings were universally set to minimum, with a resolution of 1280 x 720 (720p) across ten games: CS:GO, Dota 2, Forza Horizon 4, Genshin Impact, GTA V, Grid Autosport, Minecraft, Resident Evil 5, Skyrim, and Sleeping Dogs.




Team Pandory stated: "We tested 10 Windows games using the Chuwi Gemibook XPRO Laptop. The Intel N100 CPU is also used in Beelink Mini PC units, such as the Mini S12."

Specs:

• CPU: Intel N100 Processor
• RAM: 8 GB Of DDR5 Single Channel
• GPU: Intel UHD Graphics Xe 24 EUs @ 750 MHz
• OS: Windows 11 HOME 64-bit

View at TechPowerUp Main Site | Source

 

[kondamin]

Would this chip benefit from dual channel memory if it ever had gotten it?

I thought it weird when I saw all the boards only offering single channel

 

[LabRat 891]

Makes me miss Centaur Technologies.
VIA used to be the king of 'efficiency, at any performance-cost'.

Would be neat to see some of these ultra low power chips compared to top-end CPUs of several generations back.

 

[tabascosauz]

Arc Alchemist offers roughly double the performance when compared to the Xe-LP iGPU

Yeah, double the performance of 80-96EU Xe-LP maybe. After being handicapped by single channel (dual channel DDR5) DDR5-4800 the fact that this even hits 30fps in anything at any resolution is an achievement in itself. Though, 10-11W CPU metrics in Genshin..........isn't that kinda shit on efficiency?

E-cores compare well to Skylake so no problems there combined with a decent SSD. Stacks up pretty well against the Mendocino Athlons; maybe not GPU-wise, but if you handicap the RAM equally badly on Mendocino, who knows?

 

[sLowEnd]

Would this chip benefit from dual channel memory if it ever had gotten it?

I thought it weird when I saw all the boards only offering single channel

It’s not a board problem. The actual chip is limited by Intel to have only one memory channel.

 

[watzupken]

Would this chip benefit from dual channel memory if it ever had gotten it?

I thought it weird when I saw all the boards only offering single channel

I believe it should, especially for variants that uses DDR4 3200 memory. However, I don't expect it to perform miracles because you can get an indication by looking at the performance of the UHD 770 iGPU performance even with DDR5 memory.

 

[Nattsun]

Yeah, double the performance of 80-96EU Xe-LP maybe. After being handicapped by single channel (dual channel DDR5) DDR5-4800 the fact that this even hits 30fps in anything at any resolution is an achievement in itself. Though, 10-11W CPU metrics in Genshin..........isn't that kinda shit on efficiency?

E-cores compare well to Skylake so no problems there combined with a decent SSD. Stacks up pretty well against the Mendocino Athlons; maybe not GPU-wise, but if you handicap the RAM equally badly on Mendocino, who knows?

There's no way a cluster of 4 E-Cores at ~3GHz is consuming close to 15W on some instances... right? That's almost as much as the Steam Deck's entire Van Gogh APU!

 

[Six_Times]

Things will get interesting when nuc small form factors get Intel Ultra (mentor lake) with usb 4.0.

 

[Lew Zealand]

There's no way a cluster of 4 E-Cores at ~3GHz is consuming close to 15W on some instances... right? That's almost as much as the Steam Deck's entire Van Gogh APU!

4 E-cores consume 31W at 3.9 GHz in POV-Ray so 15W at 3.4 GHz seems plausible:

Remember: E-cores are not energy efficient for constant loads, they are die space efficient.

 

[AusWolf]

Would this chip benefit from dual channel memory if it ever had gotten it?

I thought it weird when I saw all the boards only offering single channel

That can easily be tested with any desktop CPU that has an Xe LP iGPU. I happen to have one (i7-11700), and I'm tempted to take a look when I'm back from my holiday. Although, it has 32 EUs instead of 24, but this is the closest we can get, imo.

Things will get interesting when nuc small form factors get Intel Ultra (mentor lake) with usb 4.0.

No more Intel NUCs, unfortunately (link).

 

[lexluthermiester]

Would this chip benefit from dual channel memory if it ever had gotten it?

Short answer: Absolutely yes!

 

[watzupken]

4 E-cores consume 31W at 3.9 GHz in POV-Ray so 15W at 3.4 GHz seems plausible:

Remember: E-cores are not energy efficient for constant loads, they are die space efficient.

Ironically, the E-cores were never meant to be efficient to begin with. At least I don't think it was Intel's plan to make it efficient. They are just smaller cores to make up for the lack of performance cores so that they can keep up with AMD and ARM chips. The reason why I don't think Intel ever intended for these cores to be "efficient" is because you don't need 16 E-cores to run less demanding applications. They are just stuffed in the chip to keep up with the 16c/32t AMD chips. That's why as an Alder Lake user, the current Raptor Lake and its refresh are not good in value to me. They are very powerful x86 chips, but at the same time, they draw exponentially more power with each iteration. Also, Intel is asking for quite a bit more money for a slight bump in cache, and dumping of cheap cores, instead of giving people more cutting edge P-cores.

 

[AusWolf]

Ironically, the E-cores were never meant to be efficient to begin with. At least I don't think it was Intel's plan to make it efficient. They are just smaller cores to make up for the lack of performance cores so that they can keep up with AMD and ARM chips. The reason why I don't think Intel ever intended for these cores to be "efficient" is because you don't need 16 E-cores to run less demanding applications. They are just stuffed in the chip to keep up with the 16c/32t AMD chips. That's why as an Alder Lake user, the current Raptor Lake and its refresh are not good in value to me. They are very powerful x86 chips, but at the same time, they draw exponentially more power with each iteration. Also, Intel is asking for quite a bit more money for a slight bump in cache, and dumping of cheap cores, instead of giving people more cutting edge P-cores.

Since one P-core is the same size as 4 E-cores, I would rephrase the question "how efficient is one E-core vs one P-core" as "how efficient are four E-cores compared to one P-core".