![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Lol unless your running an all core full load doing work it's just a gross exaggeration of real world use especially gaming but hardware unboxed loves to use an all core full load benchmark or workload to say it's "hot" when realistically not what the majority of users would experience.
Yeah but supposedly a 5-10% internal yield improvement. They get to charge the same while getting better binning and better yields so profits on them go up.I see what you're saying. But at the end of the day it's a refresh of an refresh.
Granted, idle to low usage consumption is much more real workload nowadays.
Which 8 core CPU would that be ? It is quite above a 7700x, 11900k is that last intel 8 cores, if we are talking some game I am pretty sure the 7800x3d will beat the 24 core threadrippers in all of them, it is not like we should expect adding core to help for 99% of them.
If we can easily imagine a 16 core losing to a 8 core by a good amount in games:
I mean that also shows AMD’s 16 core CPUs loosing to 8.
Yes thus the easily imagine a 16 core losing to a 8 core comments.
Games don’t scale automatically, thread allocation is determined during the initial engine configuration, going beyond 6 or 8 has very few real world benefits there for existing engines.
To be fair the Intel performance per watt is competitive, with a space heater…
Why is it odd if we're comparing workloads to workloads?
Core counts are different? Why not compare single core to single core?
Comparing apples to oranges because they are both fruits.
Cpu's are not fruits. You do not eat and taste a cpu, but you use it to compute stuff. Results of said computations will be identical regardless of the underlying architecture, so the performance per core is not a relevant metric for the end user.
Everything you just said here is completely wrong. It's actually astonishing anyone on this forum could be this wrong about the subject that it's blowing my mind.
The comparison here is not cores count but FPS numbers that you can get from top CPUs, so it's fair enough.
These aren't basic math computations that are being made, and with AI coming to the forefront of compute I cannot state how inaccurate your statement is.
Performance per core is still just as relevant as the above, especially clock-for-clock compared to the previous generation for comparison.
So you're suggesting that running the same computation on either AMD or Intel will result in different results? emphy is right here, and I have no idea what you are suggesting. What does AI have to do with anything?
Performance per core isn't really relevant. AMD is using all "normal" cores, Intel is using a mix of P and E cores. You can't do 1:1 comparisons. Who knows, maybe one day desktop CPUs will look like more like ARM CPUs and have fast and big "prime" cores, performance cores and efficient cores? Or be designed even more different than that?
People often get hung up on performance per watt metrics but its largely irrelevant for desktop systems. Performance per watt came to the forefront of CPU reviews when Intel started pushing that narrative. They started doing that because Intel has always cared about its server market far more than it has the desktop market. In a data center, performance per watt matters when electric bills are in the multi-thousand dollar a month range. Getting more performance per watt and more performance per core in a given amount of rack space is compelling in such cases. Not only does electric bills come into play for data center operations but rent / floor space and how much computing power they can cram into that space all becomes relevant.
Precisely. I've never once thought about my electric bill when buying hardware. The fact of the matter is if you are gaming, a high end GPU is the bigger hit to power consumption anyway. To further that example, I've never heard anyone state that the reason they opted not to go with a higher end CPU had anything to do with power consumption. It usually boils down to what their computing needs are and if they'll benefit from extra CPU cores and how much they are willing to spend on the CPU itself. It's not because CPU A is going to raise their electric bill significantly over CPU B.
You're mostly right, but
I am not saying that there isn't a technical difference between AMD and Intel when it comes to power consumption and heat output. Obviously there is. What I'm saying is that on a high end build, it doesn't really matter in a practical sense. Where the rubber meets the road, you use pretty much the same coolers on both to get the job done. Sure, you can get away with less on the AMD side, but that's not optimal. You could very well be leaving some performance on the table by using something like the Wraith Spire. How much depends on what you are doing with it. Yes, this is a larger problem on the Intel side but it isn't as if you are going to see a lot of high end AMD CPU's being cooled with a Wraith Spire. Most people will at least get a decent high end air cooler or a decent AIO for something like that. Again, these are solutions you could use on Intel CPU's just as easily.
"Pretty well" is oblivious lie..
It’s that cache it makes up for the bottlenecks that 2 memory channels imposes on gaming workloads. But neither AMD nor Intel can stretch out to 4 on consumer devices because that eats into the workstation market and they can’t have that.
I didn't disagree with this on a technical level. I disagree on a practical level. And yes, a Wraith Spire over a 360AIO or custom water is sub-optimal. Sure, it works but it doesn't work as well. Back in the days of fixed CPU clocks you'd be right. However, in today's systems keeping your CPU cooler allows it to achieve higher boost clocks and maintain them longer. Granted, cooling isn't the only condition that governs this but the point stands.
I don't know where you are getting this from. How would you be restricted by it? How would you have to compromise? I've built lots of systems with these CPU's and I've never found this to be the case. What you are talking about is theoretical, not practical. It's damn sure not the way it really works. You can run RAM at the same speeds, you can still use the same SSD's without throttling, you can still get the same performance out of the GPU's between both platforms. So how are you going to be restricted by a 14900K over a 7950X3D?
Again, the high end AMD CPU's do not have a TDP of 90w and even if AMD reported that, it would be bullshit. Those CPU's have pulled upwards of 160w or so under testing. I suspect you are talking about the 7800X3D, but that's not AMD's only CPU and gaming workloads are only part of the equation. Some people use their machines for more than gaming.
I'm not sure where you are getting this stuff from. If your case has proper airflow you aren't dumping all that into the case. Most of it should get vented out. Sure, your internal system temperature might be slightly higher but it doesn't matter in a practical sense. It's not limiting your GPU or your RAM in some way. Again, I have no idea where you are getting this stuff from.
Well, I'm comparing chips in closer price points. The 14900K and 13900K should primarily be compared to the 7900X, 7950X and their X3D counterparts. PC Perspective showed the 7950X3D pulling 150w under testing.
In theory, cheaper motherboards with lesser VRM's will degrade faster but still last sufficiently long enough to make it through their expected service life with a comfortable margin beyond that. Of course high end boards with even more vastly overbuilt VRM's will not run as hot as they spread the workload out and run at a range that's far more efficient. That's why the dreaded MSI X570-A was classified as a non-starter for overclocked 12 core CPU's and a no go for 16 core CPU's. It still works, but who knows how long the board would last doing it. I've run 12th and 13th generation CPU's on B760 boards. It's fine.
No, I can't say that with 100% certainty. I haven't tested or evaluated every budget Intel board in existence but yes, you can run those CPU's on some pretty cheap motherboards. I've personally run 13700K's and 12900K's on some pretty cheap boards like the B760 Steel Legend from ASRock which I reviewed here. Tom's Hardware reviewed a B750 motherboard with a 13900K here.
Yes, linked above.
I'm pretty confident you can run a 13900K or 14900K on most boards without issue or performance loss. I've never seen any motherboard that couldn't do it. Though that's not to say that such boards don't exist.
Both AMD and Intel get a little squirrely when dealing with boards running the B620M or H610 chipsets respectively, some of the really cheap ones there max out at 65W power delivery, whereas others will give AMD the full 175w or Intel the full 150W and others still may or may not have access to the higher power steppings, which the 7950x can top out at 290W and the 13900k 253W, though the situations that create those draws don't exist for long outside of test benches.
I've never tested on less than B650 or B760 chipset based boards. Using a $600+ CPU on a H610 chipset seems unlikely. Though your point is taken. Such boards probably wouldn't handle those CPU's all that well.
Again, that's a 7800X3D and gaming isn't the only application people use their systems for. You can load up cores doing other things besides benchmarking. If you really push an AMD CPU and you are using PBO, etc. you are going past that TDP. Period. Yes, Intel CPU's pull more power. No one is saying otherwise.
Technically, I suppose. However, on the high end people who are interested in the top offerings from either camp are likely to buy the best air coolers, better AIO's or go to watercooling. They aren't likely to stick with a Wraith Spire or put the cheapest cooler possible on a 14900K. Coolers which can likely handle either AMD's 7950X or Intel's Core i9 13900KS/14900KS, etc.
Outside of SFF this is rarely an issue.
You are splitting hairs. A good quality case with a good airflow design can handle Intel and AMD systems just fine.
While my board reviews are conducted this way, I still have personal systems in full cases. I've also built a number of AMD and Intel systems with mid-range to high end CPU's and GPU's in cases. I've probably built hundreds of machines professionally and personally over the last 28 or 29 years. This also includes multi-processor workstations and servers.
Again, I've never seen the heat difference by going to Intel CPU's negatively impact GPU clocks or memory clocks in a significant or measurable way. Going from a 10980XE to a 12900K and keeping the same GPU didn't mean my GPU suddenly got faster or had better clocks as a result. The 10980XE package power was around 500w if I recall correctly when overclocked to 4.8 or 4.9GHz. I've also run multi-CPU and GPU systems in cases. Again, what you are saying is conceptually true. (Theoretical wasn't the right word.) Not practically true. It's not that you are wrong, the heat is there but it isn't enough in a well built system to create the limitations you are talking about.
Again, gaming isn't the only consideration.
I've never tested boards that cheap. Being a reviewer for an enthusiast site, there isn't much appetite for H610 boards and things like that. That being said, Intel boards have historically been slightly more expensive than their AMD counterparts anyway.
The 7800x3D full draw goes up to 160w plus, but that’s still a lot less than the 13900k at 250 plus no not half but still significantly less.
