Support NeoGAF

[winjer]

The curious thing is that on mobile, some companies are doing away with the E-Cores, while getting a big improvement to both performance and power efficiency.

 

[kikkis]

The curious thing is that on mobile, some companies are doing away with the E-Cores, while getting a big improvement to both performance and power efficiency.

Apple calls its small core efficiency cores, but they are middle cores really. Android has been moving to same direction by ditching the efficiency cores in favor of middle cores.

 

[winjer]

Apple calls its small core efficiency cores, but they are middle cores really. Android has been moving to same direction by ditching the efficiency cores in favor of middle cores.

It's also what AMD has done with the Zen C cores.

 

[SonGoku]

Having Efficient and Performance cores do make sense if you have a battery and want to reduce power consumption. For desktop, just have Performance cores and downclock/overclock them on the fly. Not sure what Intel is trying to do here. Or maybe is about wafer-transistors; for general computing, it's better to have 4 Pcores+4 Ecores than six Pcores. For gaming, it sucks.

Intel E cores are not about reducing power consumption, they are about maximizing die area performance for heavily multithreaded apps/tasks

 

[Celcius]

285K actually competing with a 7800X3D in games after some tweaking?



Looks like CU-DIMMs can make a huge difference depending on the game
He teases at the end that the 9800X3D is going to destroy everything else and he's testing it now though...

 

[nowhat]

285K actually competing with a 7800X3D in games after some tweaking?

I didn't know Leonidas had an alt.

(didn't watch the video, and yes, I'm kidding)

 

[simpatico]

The price of the 7800X3D is getting a bit silly. It’s almost €500 here in NL from a low a few months back of €319.

Holy crap. I think I paid $350 for a cpu mobo and ram bundle a couple months ago. I'm looking at the 285 as an almost fixed 1300k. Power consumption looks better. At least it doesn't grenade itself! Unless something has come to light...

 

[Sophist]

Intel E cores are not about reducing power consumption, they are about maximizing die area performance for heavily multithreaded apps/tasks

The strategy doesn't seem to be successful. 285K has 8P+16E for a total of 24 cores but only 24 threads because of no SMT. 9950x has only 16 cores but 32 threads because of SMT. From what I have seen here and there, the 285k is barely faster for heavily multithreaded tasks like encoding or rendering. It's behind on Blender 3d.

 

[ap_puff]

285K actually competing with a 7800X3D in games after some tweaking?



Looks like CU-DIMMs can make a huge difference depending on the game
He teases at the end that the 9800X3D is going to destroy everything else and he's testing it now though...

Didn't most reviewers get cu-dimms to review with? That sounds like some platinum grade copium, as intel would never have let reviews go out in that state if it was simply a problem of test hardware being incompatible or less than ideal

 

[Bojji]

Didn't most reviewers get cu-dimms to review with? That sounds like some platinum grade copium, as intel would never have let reviews go out in that state if it was simply a problem of test hardware being incompatible or less than ideal

Most reviews were with super fast DDR5. His cyberpunk results are also surprising high compared to other reviews.

 

[ap_puff]

Most reviews were with super fast DDR5. His cyberpunk results are also surprising high compared to other reviews.

"We found the missing performance guise!"

*Edit* I rewatched his earlier review, the test system wasnt even his, it was a prebuilt system. Not really sure what's going on with that as we don't know what tweaks were made to the BIOS.

 

[Celcius]

I'd love to see someone testing this 9600mhz g.skill kit:

https://www.amazon.com/dp/B0DL1HRWDR/ref=twister_B0DL1LHFQ3?_encoding=UTF8&psc=1

(ignore the scalper pricing)

I haven't seen anyone else testing BL3 but that game stood out the most to me with regards to RAM bandwidth making a difference for AL.

 

[winjer]

Intel Core Ultra 200K review summary shows -6% gaming, +5% applications performance vs. 14th Gen - VideoCardz.com

Core Ultra 200K reviews combined For those who haven’t had time to read the Intel Arrow Lake CPU reviews and prefer not to rely on a single review, 3DCenter has compiled this information. Much like our own review roundups, they focus on checking data from multiple reviews, but they go a step...

videocardz.com

 

[Bojji]

Intel Core Ultra 200K review summary shows -6% gaming, +5% applications performance vs. 14th Gen - VideoCardz.com

Core Ultra 200K reviews combined For those who haven’t had time to read the Intel Arrow Lake CPU reviews and prefer not to rely on a single review, 3DCenter has compiled this information. Much like our own review roundups, they focus on checking data from multiple reviews, but they go a step...

videocardz.com

245k is POS, lack of HT? 6 performance cores without HT is not... much.

 

[FireFly]

245k is POS, lack of HT? 6 performance cores without HT is not... much.

It doesn't seem to be doing straightforwardly worse in heavily threaded benchmarks. For example the rendering results are decent. No HT but those E cores should be around Raptor Lake IPC, so it balances out.

 

[DonkeyPunchJr]

Intel hasn't sold a single Arrow Lake CPU at Germany's largest retailer — Core Ultra 200S sales stagnate after just one week

Arrow Lake's disappointing sales match its underwhelming performance.

www.tomshardware.com

 

[dave_d]

Intel hasn't sold a single Arrow Lake CPU at Germany's largest retailer — Core Ultra 200S sales stagnate after just one week

Arrow Lake's disappointing sales match its underwhelming performance.

www.tomshardware.com

Techepiphany, doing the work that tech journalists won't.(Like actually asking people how many units have you actually sold.)

 

[SonGoku]

The strategy doesn't seem to be successful. 285K has 8P+16E for a total of 24 cores but only 24 threads because of no SMT. 9950x has only 16 cores but 32 threads because of SMT. From what I have seen here and there, the 285k is barely faster for heavily multithreaded tasks like encoding or rendering. It's behind on Blender 3d.

Its not in my interest to defend Intel, just pointing out that E cores just like AMD c cores are optimized for die area efficiency, to getting the most performance out of each mm2 of die area. They are not more power efficient cores and where not designed for that purpose

It doesn't seem to be doing straightforwardly worse in heavily threaded benchmarks. For example the rendering results are decent. No HT but those E cores should be around Raptor Lake IPC, so it balances out.

Video games CPU threads are latency sensitive so going from P to E cores causes performance penalty, same reason why going from one CCD to the other causes a performance penalty in gaming. Something like the 9900X only has 6 cores usable for gaming despite having 12 total across two CCDs

 

[SonGoku]

Delete DP

 

[marquimvfs]

Or maybe that's just a soft launch, with very limited stock. It happened before. A few processors for few stores just to say that it arrived.

Intel Core Ultra 200K review summary shows -6% gaming, +5% applications performance vs. 14th Gen - VideoCardz.com

Core Ultra 200K reviews combined For those who haven’t had time to read the Intel Arrow Lake CPU reviews and prefer not to rely on a single review, 3DCenter has compiled this information. Much like our own review roundups, they focus on checking data from multiple reviews, but they go a step...

videocardz.com

Paper launch like I said, according to the same source:

 

[Celcius]

Some good discussion on cpu architecture

 

[winjer]