Rootax
Veteran
And that's the problem he's talking about. On paper you should not go above 375 with only 2*8pin pcie-e (+ the slot itself).
Power consumption for GPUs
- Thread starter DavidGraham
- Start date
And that's the problem he's talking about. On paper you should not go above 375 with only 2*8pin pcie-e (+ the slot itself).
Nebuchadnezzar
Legend
What does the software reporting (which is accurate) have anything to do with this? AMD lets you set 115% power limit on a 355W card. Besides the talked about cards few posts ago had 3 power connectors, only the reference has 2, and they're running >355W out of box.
The 150W limit is just for bargain bin chinesium cables, most things out there support around 288W.
conclusion to this discussion is that you are not reliable source of information.
DavidGraham
Veteran
No, the conclusion is that a lot of people need to educate themselves more on power figures, so repeat after me: some Youtube video with software overlay: not accurate source on AMD GPUs, you need to add more wattage to reach the correct number, otherwise you are just spreading unreliable misleading numbers.
Oh and once more, the new AMD sensor is still not a 100% accurate. In this shot, there is a 7w difference between the TBP sensor and the PMD measurement.
Last edited:
Your original claim was that all YouTube videos have incorrect consumption for AMD, while in reality some (as you finally noted too) might have overlay with incorrect numbers.
There's nothing indicating AMD numbers aren't accurate, they just measure it after VRM losses instead of before, which easily explains 7W difference.
Nebuchadnezzar
Legend
This is also just wrong ... any difference above tweet is simply mismatching samples rates or simply insertion loss of the cheap PMD.
DavidGraham
Veteran
I didn't imply that, so note the following:
All Youtube videos of pre 7000 series are inaccurate. As those measure TGP not TBP.
All Youtube videos post 7000 series are still inaccurate, as 99% of Youtubers are clueless, they won't use the latest software, and they won't use the new TBP data, they'll just use the old tried and tested (and wrong) TGP data. The audience will need to check the software version with the author of the video, and also make sure he is probing the right sensor, they could also try to educate him about it, or just simply give up.
Maybe after some months have passed when the major software overlays change their default behavior in new stable versions (instead of the Beta ones) that prove to be popular among users, maybe then these videos would be a lot more reliable. But right now they are not.
Your only source of reliable power figures is from Review sites/channels doing proper power testing with external tools.
The only one spreading misinformation and fud about AMD GPU's -- as usual -- is your side of the GPU politics aisle. We've already had to sift through your numerous "you have to add 50w to the power figure for 7900 XTX" posts in the review thread.
If we were discussing the power characteristics of Nvidia GPUs no doubt you would be posting something more sensible and -- dare I say it -- measured.
It's lamentable that we have to deal with people with such clear biases on this forum. I hope they're at least a shareholder or employee. Then at least we could rationalise this behaviour.
Articles: TechPowerUp is probably my go-to. Arstechnica occasionally put out good reviews. It's sad that both Anandtech and The Tech Report are no more, as no site has really stepped up to their level of detail.
Videos: GamersNexus easily is the most thorough and reliable. LinusTechTips have good reviews and bad reviews, but they generally do a decent job. Hardware Unboxed are good too. Ignore the Nvidia fanboys with a hate-boner for GN and HUB -- they're just irked that the reviewers didn't merely parrot Nvidia PR lines. They're equally harsh with AMD (and Intel), but unfortunately selection bias blinds them from that inconvenient truth.
DavidGraham
Veteran
Actually no, people were happily sharing Youtube vidoes here on the forums with the wrong power values, they were comparing them to other GPUs and reaching the wrong conclusions. Most people didn't and don't know about this power reading disparity on AMD GPUs, I certainly didn't. But now they know.
Guess what? That points still stands, and is 100% correct, because none of these Youtube videos posted so far are using the correct sensor data, and they are still at odds with the power figures obtained from adequate measurements made by professional reviews. You cannot use Youtube videos reliably for power figured from AMD till this day.
I never saw you correct any of the wrong power values posted here, you only posted to add noise to a purly technial discussion. In otherwords, you don't care about the truth, you are indulged in your own self bias that you accuse me of, which is fine, just try to stick to discussing the technical points at hand without resorting to low personal attacks.
Going personal is a weakness, cant debate the actuall matter so lets go personal…..
Its tiresome and should be moderated anytime it happens.
Most people here know better than to trust the word of random no-name YouTubers. You being duped or putting any worth in their gossip is your own folly.
If you’re so concerned with misinformation, why stop at YouTube even? Why don’t you tackle nonsense on TikTok, Twitter, and reddit too.
No, I don’t get personal. And I don’t add fuel to the fire by using random YouTube channels. I just read, listen, and wait and stick to trusted review sites with open testing procedures, accountability, and that listen to public feedback.
Man from Atlantis
Veteran
Similar threads
- Deleted member 13524
- Architecture and Products
