Follow Tom's Hardware on Google News , or add us as a preferred source , to get our latest news, analysis, & reviews in your feeds.
Zak Killian Contributor Zak is a freelance contributor to Tom's Hardware with decades of PC benchmarking experience who has also written for HotHardware and The Tech Report. A modern-day Renaissance man, he may not be an expert on anything, but he knows just a little about nearly everything.
hotaru251 i mean….thats fine? You wouldnt want to run a cpu past 90c long term anyways. Reply
TerryLaze Ok! So? Well hopefully this will keep you guys from running intel CPUs insanely out of spec in your benchmarks…although I already see you guys limiting intel to that 5.1 and claiming it's "out of the box" ,man, we are just reviewers man, we can't set up a decent system… I hope it doesn't just limit the clocks but actually turns all the settings to the default. hotaru251 said: i mean….thats fine? You wouldnt want to run a cpu past 90c long term anyways. This will let you run it at whatever the tjmax is (prob. 105) forever, just not at full tilt. Reply
ATIRAGEPROTURBO TerryLaze said: Well hopefully this will keep you guys from running intel CPUs insanely out of spec in your benchmarks…although I already see you guys limiting intel to that 5.1 and claiming it's "out of the box" ,man, we are just reviewers man, we can't set up a decent system… I hope it doesn't just limit the clocks but actually turns all the settings to the default. This will let you run it at whatever the tjmax is (prob. 105) forever, just not at full tilt. Tell me you didn't read the post without telling me you didn't read the post. Reply
usertests However, this 5.1 GHz limit could be a downer for people who are interested in the rumored 52-core version of Nova Lake. There were always going to be thermal/power challenges with loading up the entire full 52-core monster. This is probably going to be acceptable. For gaming, very few of those cores are actually needed outside of a lightning fast shader compilation (which could be mitigated by precompilation). Reply
thestryker The only thing which would make this concerning to me is if 5.1GHz was the stock all-core clock, but I cannot imagine this being the case. To me what this really looks like is a way to push people towards keeping stock power limits and/or ensuring good cooling. At stock power limits ARL is pretty easy to cool and I never saw 90C on either 265K or 270K Plus with them in place (though encoding is the only heavily multithreaded thing I do). Assuming NVL is similar cooling wise (I'd bet bLLC will be easier to cool than non) this limit should mean nothing for the majority of users. Reply
bit_user usertests said: There were always going to be thermal/power challenges with loading up the entire full 52-core monster. This is probably going to be acceptable. Pfft. I'd like to see how many watts a dual-tile Nova Lake burns, even running all its P-cores at 5.1 GHz! I think just to get temps and power consumption down to reasonable levels, it'll be necessary to reduce clock speeds even more than 5.1 GHz. Reply
bit_user thestryker said: At stock power limits ARL is pretty easy to cool and I never saw 90C on either 265K or 270K Plus with them in place (though encoding is the only heavily multithreaded thing I do). Have you run a traditional multithreaded benchmark, like Cinebench or SuperPi? What's your all-core clock speed, power, and thermals, if you do this on just the P-cores? Reply
TerryLaze bit_user said: Pfft. I'd like to see how many watts a dual-tile Nova Lake burns, even running all its P-cores at 5.1 GHz! I think just to get temps and power consumption down to reasonable levels, it'll be necessary to reduce clock speeds even more than 5.1 GHz. What are the reasonable levels for a potential 52 core desktop CPU running heavy loads at 5.1Ghz??? We are talking about something that never existed before…so how do we set realistic "reasonable" levels? This will potentially be more than twice the cores of the next halo ryzen cpu… Reply
bit_user TerryLaze said: What are the reasonable levels for a potential 52 core desktop CPU running heavy loads at 5.1Ghz??? I guess I just mean what you can actually dissipate, at steady-state. TerryLaze said: We are talking about something that never existed before…so how do we set realistic "reasonable" levels? In real-world, multithreaded workloads, it won't be just the P-cores that are active. However, we could look at what a CPU like the 9950X can do in all-core workloads to get an idea of what 16 P-cores could draw. According to SatterBencher #81: When running the OCCT CPU AVX2 Stability Test, the average CPU core effective clock is 4885 MHz with 1.026 volts. The average CPU temperature is 95.1 degrees Celsius. The average CPU package power is 236.3 watts. That's not a great point of reference, because: My above point that almost nobody running all-core workloads is going to limit it to just the P-cores. Nova Lake will be on a better node. The above is highly-tuned, far from out-of-the-box performance (which is "4346 MHz with 0.988 volts; package temp: 77.5 degrees C; package power: 181.4 watts"). Intel's P-cores are much larger and (at higher frequencies) more power-hungry than corresponding Zen cores. So, there are arguments going both ways, but once you add in the E-cores, I'd bet pushing the P-cores above 5.1 GHz is going to exceed what even a fairly high-end water cooling setup could dissipate. Reply
thestryker bit_user said: Have you run a traditional multithreaded benchmark, like Cinebench or SuperPi? Yeah Cinebench 2024 pulls 265-285W (no power limit) with P-cores running 5.4GHz and E-cores running 4.8GHz with peak temp 88C. I wouldn't check OCCT Extreme/SuperPI without power limits though because those are likely to peg my cooling and pull somewhere around 350W. These would be more likely to go 90C+ as well, but I'm not 100% sure since I've never used them with power limits as I've only used them for maximum cooling testing :ROFLMAO: bit_user said: I think just to get temps and power consumption down to reasonable levels, it'll be necessary to reduce clock speeds even more than 5.1 GHz. I think it would depend on workload more than anything else. I'd bet the stock clocks wouldn't be any lower than 5.1GHz for the dual Compute Tile part. The most recent leaks have seemingly held steady on PL2 at 474W for dual Compute Tile and 296W for bLLC single Compute Tile. So it'll depend on how Intel approaches boost clocks: all-core could be lower, or they could have varied all core depending on the Tile similarly to how AMD does dual CCD. Reply
Key considerations
- Investor positioning can change fast
- Volatility remains possible near catalysts
- Macro rates and liquidity can dominate flows
Reference reading
- https://www.tomshardware.com/pc-components/cpus/SPONSORED_LINK_URL
- https://www.tomshardware.com/pc-components/cpus/intel-leaker-claims-nova-lake-desktop-chips-have-5-1-ghz-clock-limit-beyond-90-c-high-clock-speeds-could-demand-significant-cooling-for-next-gen-chips#main
- https://www.tomshardware.com/membership
- Apple faces unprecedented memory cost increases as RAM prices continue to surge, manufacturing bill climbs by $33 billion
- Thermal Grizzly launches PhaseSheet PTM Pro with improved thermal conductivity
- Grab a Nintendo Switch 2 controller upgrade from $49, just in time for Zelda: Ocarina of Time's remaster
- Upgrade to an OLED display for only $269 at Walmart — Acer's 27-inch Nitro gaming monitor has a superfast 240Hz refresh rate and crisp WQHD resolution
- Slice Into GFN Thursday: ‘Phantom Blade Zero’ Launches on GeForce NOW
Informational only. No financial advice. Do your own research.