appiah4 wrote on 2022-11-07, 06:16:

TrashPanda wrote on 2022-11-07, 06:14:

appiah4 wrote on 2022-11-07, 06:11:

No, there is no MCM for nVidia next gen.

I haven't heard of them ditching their Hopper MCM designs unless something changed.

Hopper is monolithic, Blackwell will be MCM. See above.

Ahh I see, last I read Hopper was their MCM attempt, well I doubt they can keep brut forcing at any rate, no one wants to buy a 1300 watt power supply just to run a GPU or run a AC unit just to cool it.

-Turns out I was looking in the wrong place ...fuck Hopper is a behemoth ..80billion transistors and 30 VRM units just to power it ..WTH

[Melting intensifies]

Its fun to watch all the AMD fans boys melt down every release.

I was just having a bit of fun with all of the outrageous and non scientific justification posts on here. Buy what you like, I am 100% happy with my 4090, because I buy the best. I don't need to justify my purchase based on someone else's criteria they use to justify not buying something because of cost. You all could just say its too expensive for me, you don't have to make the item you are eyeing the best buy by any measure. Like I said I am happy with buying the best, no need to justify anything.

Shagittarius wrote on 2022-11-07, 13:58:

Its fun to watch all the AMD fans boys melt down every release.

I was just having a bit of fun with all of the outrageous and non scientific justification posts on here. Buy what you like, I am 100% happy with my 4090, because I buy the best. I don't need to justify my purchase based on someone else's criteria they use to justify not buying something because of cost. You all could just say its too expensive for me, you don't have to make the item you are eyeing the best buy by any measure. Like I said I am happy with buying the best, no need to justify anything.

You sure sound like you are very happy. So happy that you are trying to prove it to us. And to justify your purchase. Like every buyer's remorse story I ever heard. I'm so happy for you. Be happy, try not to burn your house down. 😀

appiah4 wrote on 2022-11-07, 14:47:

Shagittarius wrote on 2022-11-07, 13:58:

Its fun to watch all the AMD fans boys melt down every release.

I was just having a bit of fun with all of the outrageous and non scientific justification posts on here. Buy what you like, I am 100% happy with my 4090, because I buy the best. I don't need to justify my purchase based on someone else's criteria they use to justify not buying something because of cost. You all could just say its too expensive for me, you don't have to make the item you are eyeing the best buy by any measure. Like I said I am happy with buying the best, no need to justify anything.

You sure sound like you are very happy. So happy that you are trying to prove it to us. And to justify your purchase. Like every buyer's remorse story I ever heard. I'm so happy for you. Be happy, try not to burn your house down. 😀

I will thanks!

Only Conan knows what is best, for everyone else it's relative.

DosFreak wrote on 2022-11-07, 16:34:

Only Conan knows what is best, for everyone else it's relative.

That is good.

Earlier this year I was reading about melted 8-pin PCIe cables. It's common now to have cables with dual 8-pin connections running into a modular power supply via one connector. Sometimes the PSU connector can't handle the load.

I think it is all a bit too early to say yay or nay on the AMD cards.
Until we see some real world independent benchmarks, then the jury is still out.

Even then, we don't know what driver improvements and developer optimizations might come int time, from either company.
I expect that on the top end nVidia will hold the raster crown in the 10% range or so. While the RT performance will likely be much more, like 25-40% depending on the title.

All of this while costing a very generous amount of money more.
Unless you care very very much about RTX performance or think DLSS is worth the cost.......4090/4080 just doesn't make sense, to most people.
Grain of salt.

swaaye wrote:

Earlier this year I was reading about melted 8-pin PCIe cables. It's common now to have cables with dual 8-pin connections running into a modular power supply via one connector. Sometimes the PSU connector can't handle the load.

Most of the time it's user error (poor contact) or load exceeds 300W (hello, miners and "extreme" overclockers), which was measured as a safe limit for cables with at least 18 AWG (literally any half-decent PSU). But of course, having 100% headroom on both ends, to avoid potential issues, is safer. 12VHPWR cable does not have such headroom and presumably tolerance for user error (see video posted above) is much more slimmer.

Oh, you also could melt these dual 8-pin cables with Radeon 295x2, which has two connectors, but draw more than 150W from each.

Jasin Natael wrote on 2022-11-07, 19:14:

I think it is all a bit too early to say yay or nay on the AMD cards. Until we see some real world independent benchmarks, then […]
Show full quote

I think it is all a bit too early to say yay or nay on the AMD cards.
Until we see some real world independent benchmarks, then the jury is still out.

Even then, we don't know what driver improvements and developer optimizations might come int time, from either company.
I expect that on the top end nVidia will hold the raster crown in the 10% range or so. While the RT performance will likely be much more, like 25-40% depending on the title.

All of this while costing a very generous amount of money more.
Unless you care very very much about RTX performance or think DLSS is worth the cost.......4090/4080 just doesn't make sense, to most people.
Grain of salt.

To me RTX Remix is huge, making all old games sandboxes to play with, but for that I'd go with a used economical 3080 ti non arson edition.

The Serpent Rider wrote on 2022-11-07, 20:32:

swaaye wrote:

Earlier this year I was reading about melted 8-pin PCIe cables. It's common now to have cables with dual 8-pin connections running into a modular power supply via one connector. Sometimes the PSU connector can't handle the load.

Most of the time it's user error (poor contact) or load exceeds 300W (hello, miners and "extreme" overclockers), which was measured as a safe limit for cables with at least 18 AWG (literally any half-decent PSU). But of course, having 100% headroom on both ends, to avoid potential issues, is safer. 12VHPWR cable does not have such headroom and presumably tolerance for user error (see video posted above) is much more slimmer.

Oh, you also could melt these dual 8-pin cables with Radeon 295x2, which has two connectors, but draw more than 150W from each.

These cables are actually rated up to 300W each, not 150W. There is huge safety margin with these. The issue with 295x2 melting was (IIRC), again, due to poor 2x6 to 1x8 adapters.

appiah4 wrote:

These cables are actually rated up to 300W each, not 150W.

You've missed the point.

The Serpent Rider wrote on 2022-11-08, 08:19:

appiah4 wrote:

These cables are actually rated up to 300W each, not 150W.

You've missed the point.

Just re-read and yeah, I have. Sorry.

Shagittarius wrote on 2022-11-07, 13:58:

I am 100% happy with my 4090, because I buy the best. I don't need to justify my purchase based on someone else's criteria they use to justify not buying something because of cost. You all could just say its too expensive for me, you don't have to make the item you are eyeing the best buy by any measure. Like I said I am happy with buying the best, no need to justify anything.

I, for one, can certainly understand this. Some people just buy the best and that's perfectly fine.
Now, whether or not ALL of these people can ACTUALLY afford to buy the best... is a completely different matter and it's subjective, everyone has different thresholds (+ it's everyone's business how they manage their finances, anyway).

For example, to me, based on where I live, buying the best would only be a viable option for people who, on average, are able to easily save at least $10k+ per month (not talking about disposable income, I'm talking about the actual money you're left with AFTER you've paid for absolutely all essential AND non-essential/discretionary expenses). Since I've yet to hit this target, then I will settle for midrange, thank you very much. 😀

Now, let's assume that I am in that category, ergo (in my eyes) I could easily afford to buy 'the best'. Would I run and buy an RTX 4090 (hoping that I could actually find one)?
Well... very hard to say, because this is again subjective. As others have pointed out, the RTX 4090 seems more like a Pyrrhic victory than anything else. In fact, to me, all latest hardware releases from AMD, Intel and nVIDIA are... a tad disappointing: high power consumption, high temperatures (and in the case of graphics cards... insane weight and sizes as well), just to be able to claim that 3 - 5% margin that might propel them to the top of (most) performance charts (and, at the same time, giving us all kinds of "eco modes", when the vast majority of users aren't tech savvy). Can I consider any of these "the best"? No, don't think so, not when I remember past releases that I was actually excited about, that I actually felt like they WERE the best. Does anyone remember releases such as the Athlon 64, the Core 2 Duo, the Sandy Bridge, the Radeon 9700 Pro, the GTX 1080 Ti? Those WERE exciting releases, they were actually THE BEST of their time, and for many of them this didn't just mean performance, but performance with equal OR lower power consumption AND lower temps.

So, no, I'd like to think that, even as a billionaire, it'd still be a "hard pass" for me.

1080Ti is a mistake nVidia will never make again 😁

appiah4 wrote on 2022-11-08, 13:04:

1080Ti is a mistake nVidia will never make again 😁

Yeah, not for some time.
I don't have the Ti, but I am still using my vanilla Asus ROG 1080 and while it is starting to show it's age it has been a fantastic card.

I don't see the 4000 series being in that kind of a historical place.
I think that in time they will be viewed as the new Fermi cards. On the fringe of making any kind of sense.

The new Fermi cards that run at 68c under a full load. You can say Fermi with a better cooling solution, but that hardly means anything.

Shagittarius wrote on 2022-11-08, 15:54:

The new Fermi cards that run at 68c under a full load. You can say Fermi with a better cooling solution, but that hardly means anything.

Sorry, but I strongly disagree.
The core temperature is only a very small part of what is a much bigger picture. Cooling and temperature for other components (like those on the power delivery) is just as (if not MORE) important. And this becomes a big issue with such a high power consumption that puts stress on all the circuitry, particularly when many (all) AIBs will cut corners when it comes to properly cooling all major components.

For those who don't know this youtube channel which focuses on video card repairs, I highly recommend it: https://www.youtube.com/watch?v=g-uxEpIjiPU
He usually also goes into great detail regarding cooling (or lack thereof), even for premium cards that turn out to be... not so premium.

Good point, we should stop overclocking everything STAT!