I can finally make this thread.
Phoenix is indeed RDNA3. My advice to everyone: treat the old APU rumours as being out of date.
Phoenix is indeed RDNA3. My advice to everyone: treat the old APU rumours as being out of date.
Again, you're assuming the only source of CPU load is a variable component that scales with GPU load, but that's far from the case. You have actual game logic to run... That will dominate.Reducing CPU throughput by 2 will reduce power by at least 4, and in the case of AMD by 2^2.7 since that s the power/frequency rate of the process used for PHX (P = F^2.7).
Again, you're assuming the only source of CPU load is a variable component that scales with GPU load, but that's far from the case. You have actual game logic to run... That will dominate.
Also, you're ignoring the uncore, which is a significant source of power draw. You can't lump that in with the GPU.
What numbers? The link you keep posting showed like 20W core + SoC out of 25W total. And on that matter, why do you keep trying to use that to claim driver overhead or power allocation differences between the CPUs? A weakly threaded load on 4c vs 8c explains most of the gap there...I m assuming nothing, i look at the numbers
And where are you getting that number from?The uncore on a 25W RMB is 3W-4W at most
So why don't you post any actual data where a gaming workload is only using ~5W for the CPU? Or does that data simply not exist...add 4-6W for the cores and rest go to the GPU
The link i posted show 25W for CPU + GPU + uncore, the software may say CPU but that s for the whole package, off course you didnt notice that for CPU usages varying considerably from 10% to 45% power is stable at 25W, wich mean that its the whole APU power, even a 10 year old would notice it.What numbers? The link you keep posting showed like 20W core + SoC out of 25W total. And on that matter, why do you keep trying to use that to claim driver overhead or power allocation differences between the CPUs? A weakly threaded load on 4c vs 8c explains most of the gap there...
I get those numbers from the numerous links i posted and wich wich i look at accurately, so far you re only bringing clulessness since you didnt even look at anything to try to extract numbers that are significant, actually you are just trolling all the way.And where are you getting that number from?
So why don't you post any actual data where a gaming workload is only using ~5W for the CPU? Or does that data simply not exist...
CPU usage varying between 10-45% does not mean that the CPU is also consuming 10-45% of the package power.The link i posted show 25W for CPU + GPU + uncore, the software may say CPU but that s for the whole package, off course you didnt notice that for CPU usages varying considerably from 10% to 45% power is stable at 25W, wich mean that its the whole APU power, even a 10 year old would notice it.
you forgot to blur out your steam handle with your current balance. no 50 cent games to spend your big bucks on?CPU usage varying between 10-45% does not mean that the CPU is also consuming 10-45% of the package power.
View attachment 80558
View attachment 80559
You're making rookie errors.
I know, though I can't be bothered with doxxing myself on AT forums, of all places.you forgot to blur out your steam handle with your current balance. no 50 cent games to spend your big bucks on?
lol you can change it and but you're not disliked around here. I've never liked holding money in balance it gets irritating to look at. wish you could buy a game on services like steam and it behave as a transactional service like square or that other one.I know, though I can't be bothered with doxxing myself on AT forums, of all places.
You're making two contradictory claims - that the numbers claimed to be from the CPU aren't actually CPU power, and that it's showing CPU power. As others have already pointed out, load percent isn't the same as Watts. I can't believe we actually have to point that out...off course you didnt notice that for CPU usages varying considerably from 10% to 45% power is stable at 25W
So, did you think you could lie about what an article says and get away with it because people (yourself included, clearly) are too lazy to read it? There is nothing in that article saying the CPU portion of the die consumes ~5W. At no point in the article do they break power down beyond total package power.I get those numbers from the numerous links i posted
CPU usage varying between 10-45% does not mean that the CPU is also consuming 10-45% of the package power.
View attachment 80558
View attachment 80559
You're making rookie errors.
You're making two contradictory claims - that the numbers claimed to be from the CPU aren't actually CPU power, and that it's showing CPU power. As others have already pointed out, load percent isn't the same as Watts. I can't believe we actually have to point that out...
So, did you think you could lie about what an article says and get away with it because people (yourself included, clearly) are too lazy to read it? There is nothing in that article saying the CPU portion of the die consumes ~5W. At no point in the article do they break power down beyond total package power.
If we've gotten to the point that you're literally lying about having sources and link spamming to disguise that, I have to ask, why? Is it so hard to believe a CPU needs >5W for gaming? This is an absurd hill to die on.
i find not being sober when visiting this particular board refreshing and easier on the brain than visiting sober.I'm not even sure what's being discussed here at this point.
So your nonsense where you once again insist that CPU load percentage equals watts, despite that being ridiculous. A single core will happily use >5W, and in case you weren't aware, games use more than one core. This is all basic common sense that you either lack or refuse to apply.The only liar here is you and eventually Tamz, see my answer to him
On the contrary, you insisted you had numbers, which turned out to be you lying about a link you never even clicked on, hence why you're now trying to deflect and ignore that same "source". I think it's obvious you're not engaging in good faith. What your agenda is, I can't be bothered to care.You pretend that a CPU part consume 25W at any power usage percentage, and since you cant provide any number you keep doing bla bla as a cover.
So your nonsense where you once again insist that CPU load percentage equals watts, despite that being ridiculous. A single core will happily use >5W, and in case you weren't aware, games use more than one core. This is all basic common sense that you either lack or refuse to apply.
On the contrary, you insisted you had numbers, which turned out to be you lying about a link you never even clicked on, hence why you're now trying to deflect and ignore that same "source". I think it's obvious you're not engaging in good faith. What your agenda is, I can't be bothered to care.
Power to the APU is provided by three VRM stages, controlled by a MP2845. Two of the stages are marked 8690 2823 B, while a third is marked 8690 3000 C. The VRM is possibly split into a two-stage component and a separate single-stage one. Therefore, the VRM is rather weak, but that’s not a huge deal considering the APU appears to be capped at 16 W. That power budget is flexibly allocated between the CPU and GPU. For example, a GPU-bound sequence could see the GPU pulling over 10 W, while the CPU side gets squished below base clock and draws 2-3 W. The opposite applies for a CPU-only workload.
Checking the after-action report in GHPC. This activity is mostly GPU bound, and sees the power budget dominated by the GPU running full-tilt at 1.6 GHz. In contrast, the CPU gets squished into a tiny power budget and runs at cell-phone like clocks.
Run at a fixed clockspeed or cap/disable boost.So much power wasted on those CPU cores.
Isn't the a way to manually force cap the CPU to improve GPU performance?
I don't think anyone is debating that these APUs can flexibly allocate power, but being "squished below base clocks" (i.e. <2.8GHz, and surely on only a core or so) is not really viable for modern gaming. It's a theoretical possibility, but I don't see anyone demonstrating such a scenario outside of GPU-only synthetics, as they essentially call out in that review. In actual games, it would be reasonable to expect a much more even (i.e. CPU-heavy) power allocation. There's a baseline level of CPU compute games typically need, and trying to push below that makes for a terrible experience.How long can this go on...
This has been posted more than a few times regarding the Van Gogh APU.
https://chipsandcheese.com/2023/03/05/van-gogh-amds-steam-deck-apu/
specifically
At this point I have to assume you're trolling. There's no other way someone could claim that gaming is embarrassingly parallel and involves zero CPU compute beyond rendering graphics. That is not how games work. As I said, most of the CPU compute intensity from gaming is independent of the graphics pipeline, and most of that is weekly threaded, i.e. frequency demands don't change.If there s 32-40 CUs the CPU will be required to provide less than half the thoughput of the previous case, frequency of the CPU can be halved and its power will go down by at least a 5 factor down to 20W.
I suggest watching your own video for once. It showed CPU utilization, not power budget allocation, so it will obviously differ wildly when you're comparing an 8c vs 4c. But that's enough feeding the trolls from me. If basic common sense is too much to ask for, there isn't a discussion to be had here.Now and obviously an Intel IGPU require a lot more CPU power than a RDNA GPU as proved by the linked review, that were you are completely off
If you didn't notice, In that link they compare Core i7-1165G7 vs R7 5800UEvidence is here that AMD CPU part consume less that Intel CPU part in gamings, we dont even need HVinfo, the real debate would be to find the reason, either Intel has non optimised drivers or Radeon has, by design, much less driver ovehread or both.
As already pointed in the AMD/Intel comparison there s the CPU % usage that is displayed and Intel can have more than 2x the usage displayed for the AMD APU.
Since CPU usage relate to the total available power (25W for AMD in the test) then if usage is 20% imply that the CPU part use 5W of the available power; rest go the GPU + uncore.
Intel being at 50% FI mean that the CPU part use 15W (out of the 30W available in the test), rest goes to the GPU + uncore.
Unless CPU usage goes to more than 50% most of the power is distributed to the GPU, of course that s dependent of the game tested but difference can be huge between the two CPUs, in the first time stamped link below AMD is at 22% while Intel is at 53% CPU usage for exactly the same scene, wich correlate with your own test that is btw not that representative since you did test a single case.
14% and 42% respectively :