Support NeoGAF

[rnlval]

It was a firmware bug and it's fixed. New 14900k CPUs don't have this issue. Even Zen 5 is frying itself due firmware bugs it's not exclusive to Intel.

For AM5, only ASRock has the voltage issue. This issue has been solved with a BIOS update.

 

[rnlval]

In higher core count workloads the u9 and r9 are neck and neck.
If you are buying high core count CPUs and NEED them then Intel is totally viable.
Down the stack the 12/3/400 were always good and I dont believe Intel ever brought out the replacement 240 or atleast havent seen any in the wild.

Cinebench 2024: Reviewing the Benchmark

Maxon’s Cinebench is a perennial benchmark favorite.

chipsandcheese.com

Cinebench 2025 doesn't use vector 256-bit or vector 512-bit. It's exploits scalar 3-operands floating point from AVX2. Cinebench 2025 has the old scalar 3 operands from the old school RISC workstation world.

 

[ResilientBanana]

3nm 5nm 7nm?

From what I recall I think they are aiming for 2nm.

 

[StereoVsn]

Man, this just seems to be Windows Scheduling nightmare. You just know first 6 months to a year lost release it will run terribly with games without manually turning off E cores or using stuff like lasso.

 

[SpokkX]

It's a recent thing, the devs of Doom TDA talked about how they had to optimize for heterogenous CPU architectures because they're a new recent phenomenon that is not going away.

Are you sure this is the correct way to go.. i doubt it

Arm has no such concept

 

[dgrdsv]

3nm 5nm 7nm?

Should be their 18A node but if that'll fail they'll likely use TSMC's N2.

 

[blastprocessor]

Ok if 2nm then AMDs next chip is rumoured to be the same, with 12 cores on 1 CCD.

 

[Gameplay Gods Bless]

For AM5, only ASRock has the voltage issue. This issue has been solved with a BIOS update.

It wasn't only ASRock happened on Asus and others as well but ASRock was the worst by far.

 

[Gameplay Gods Bless]

Are you sure this is the correct way to go.. i doubt it

Arm has no such concept

What? Literally modern arm cores including Apple (apples A15 cpu has 2 P cores and 4 E cores), arms own and even Qualcomms are heterogenous with different types of cores in the same cpu. Hell this trend started with ARM, if anything x86 is late to the party. The 11 year old Qualcomm SD 808 had two
Cortex-A57 cores and 4 Cortex-A53 cores.

Nvidia's new GB10 CPU used in their upcoming laptops has ten Cortex-X925 cores (P cores), ten Cortex-X725 cores (E cores).

 

[rnlval]

It wasn't only ASRock happened on Asus and others as well but ASRock was the worst by far.

For context,

"There currently have been around 100 reported cases on Reddit of 9000 series x3D failures."
From https://www.tomshardware.com/pc-com...ority-may-be-happening-on-asrock-motherboards
"Moderator SoupaSoka on the ASRock subreddit created a "9800X3D Failures/Deaths" megathread revealing an allegedly high amount of user reports surrounding dying 9800X3Ds on ASRock motherboards. Over 40 reports have been made so far, with 32 of them on ASRock's boards."

Total found now potentially 157 dead 9800X3Ds, 130 on ASRock boards, 20 ASUS, 5 MSI, 2 Gigabyte.

Case with multiple dead 9800X3Ds on the same mobo: 23 and 143.

Dead 9800X3D tally based on chipset:

• B650: 23 cases
• B850: 56 cases
• X670: 9 cases
• X870: 68 cases
• A620: 1 case

Also 16 other 9000 series CPU, 15 ASRock boards, 1 Gigabyte.

Also 8 cases for the 9950X3D, 1 ASUS, 7 ASRock, 7 on X870 chipset and 1 on B650.

There are millions of AM5 motherboards and CPUs. ASRock's result is abnormal since they are FAR above other brands' failure incident reports.

 

[Dalbajobas]

So 500W then?

It's a typo... 1,500 W on load.

 

[rnlval]

Ok if 2nm then AMDs next chip is rumoured to be the same, with 12 cores on 1 CCD.

AMD already has 16 cores Zen 5c "Turin Dense" CCX with shared L3 cache.

Zen 5c "Turin Dense" has the full 512-bit AVX hardware implementation, which is different from Strix Point's Zen 5c.

 

[winjer]

https://videocardz.com/newz/intel-nova-lake-performance-leak-claims-10-single-and-60-multi-threaded-uplift

The Core Ultra 200 series ended up being around 5% slower than the previous generation in gaming benchmarks. Nova Lake, expected to launch next year for both desktops and laptops, is rumored to deliver at least a 10% uplift in single-threaded performance (overall, not just gaming). A slide shared on X shows Intel targeting a 1.1x gain in single-threaded and 1.6x in multi-threaded performance, presumably compared to the Core Ultra 200 series.

The slide doesn't confirm which product segment this refers to. It could be a mobile chip, or a non-flagship configuration, for all we know. What's clear, though, is that Intel is once again focusing on gaming performance, something that wasn't highlighted during Arrow Lake's rollout. The slide even mentions "leadership in gaming performance," which implies a potential performance edge over AMD's Ryzen 9000X3D series, that would be a notable shift if true.

 

[Insane Metal]

Yes please. Bring on competition. +60% in MT is a pretty big deal if they manage to accomplish.

 

[KeplerL2]

Yes please. Bring on competition. +60% in MT is a pretty big deal if they manage to accomplish.

60% MT with double the cores, new P+E uarch and new node is absolutely atrocious.

 

[Insane Metal]

60% MT with double the cores, new P+E uarch and new node is absolutely atrocious.

Not the same cores, different architecture and new node, I think it's too soon to say it's atrocious... considering Core 2xx went backwards in some respects.

 

[winjer]

Intel Nova Lake CPUs may finally bring a 3D V-Cache rival to desktop gaming

According to tipster @Haze2K1, at least two SKUs in the Nova Lake lineup will ship with increased L3 cache. Intel calls the new technology "bLLC," which is...

www.techspot.com

According to tipster @Haze2K1, at least two SKUs in the Nova Lake lineup will ship with increased L3 cache. Intel calls the new technology "bLLC," which is short for "big Last Line Cache." The leaker added that both SKUs with bLLC will feature 8 P-cores and 4 LP-E cores. One will be paired with 20 E-cores, while the other will include only 12. Both chips are expected to have a 125W TDP.

 

[dgrdsv]

60% MT with double the cores, new P+E uarch and new node is absolutely atrocious.

Yeah, these numbers don't look well at all.
+10% ST perf would essentially return NVL to what they've had in RPL-S.