nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Child Item

Nvidia RTX Discussion

K.O.Kiki

A new era in 3D graphics stands before us.

First-gen Raytracing GPUs:

Nvidia "Turing" - Quadro RTX 5000 / 6000 / 8000 (US$ 2'300 / 6'300 / 10'000)

https://www.anandtech.com/show/13214/nvidia-reveals-next-gen-turing-gpu-architecture
https://www.anandtech.com/show/13217/nvidia-announces-turing-powered-quadro-rtx-family

Find more posts tagged with

Quick Links

Comments

jonski

Samuel T. Cogley wrote: »

What was the question? :pac:

If you are talking to me then the reason I didn't repeat the question was because I thought I was replying directly after the last comment relating to it

Being the clever boy that I am I missed an entire page of other comments

The original question was how would my 650w psu hold up to a 3080 .

Squidgy Black

JoyPad wrote: »

If it's listed as 294mm, it should fit, right?

Yep that one can get added to the list! Hadn't noticed OC had updated the description of any of them, couldn't find any mention of dimensions when I looked the other day.

Also interesting that there's very few dual fan versions so far. I wonder does this combined with the high TGP mean that any AIBs that don't get fan curves and bios dead on could be in for a bit of bother heat wise

Samuel T. Cogley

What was the question? :pac:

JoyPad

Squidgy Black wrote: »

I'm in a similar boat with a Corsair 280x. The only ones I've seen so far that have confirmed under 300mm for the 3080 are the founders, the EVGA XC3 and the FTW3 (although this is 300mm so might be tight in my case) and the ASUS TUF.

If it's listed as 294mm, it should fit, right?

jonski

Yes, no one else needs to say it, I'll say it myself....... I was stupid asking the question without giving more details of my system........but....I'm old so .

i7 7700K
GIGABYTE GA-Z270-HD3P 1151 7th Gen DDR4 Motherboard
Corsair CMK16GX4M2B3200C16 Vengeance LPX 16 GB (2 x 8 GB) DDR4 3200 MHz
EVGA GeForce GTX 1080 SC GAMING ACX 3.0, 8GB GDDR5X
HDD - 1x 250g ssd, 1x 500g ssd, 1 2tb sata .

No overclocking ( I know I know, stupid buying a 7700k when I had no intention of OC )

with that in mind how does my my 3 year old EVGA SuperNOVA 650 P2, 80+ PLATINUM 650W look now if I added in a 3080 .

Squidgy Black

slade_x wrote: »

It would be nice if they were supplied with card dimensions for more of the different versions they are going to offer. Im currently using a CM Dominator II, Ill only have about 305mm length so im limited to the ones that are stated at 280 to 300mm

I have an old CM Stacker full tower case but don't want to use it as its massive, and also don't fancy moving everything over to a new case just to upgrade card.

I'm in a similar boat with a Corsair 280x. The only ones I've seen so far that have confirmed under 300mm for the 3080 are the founders, the EVGA XC3 and the FTW3 (although this is 300mm so might be tight in my case) and the ASUS TUF.

Samuel T. Cogley

Jesus a CM Stacker, that takes me back.

slade_x

Mr Crispy wrote: »

OcUK main man Gibbo just posted this on their forum, re Ampere availability;

They're also limiting purchases to one per customer.

It would be nice if they were supplied with card dimensions for more of the different versions they are going to offer. Im currently using a CM Dominator II, Ill only have about 305mm length so im limited to the ones that are stated at 280 to 300mm

I have an old CM Stacker full tower case but don't want to use it as its massive, and also don't fancy moving everything over to a new case just to upgrade card.

jebidiah

BloodBath wrote: »

That's absolute peak, you're practically never going to get near that. Most good quality PSU's can handle spikes above their rating for short periods as well.

Best advice for people worried is to get a plug in power monitor. Not that expensive and pretty useful. I remember testing my rig with my Vega 56 when I got it. Said I needed a 650W PSU and I dont think I ever went over 400W with a Ryzen 1600 Overclocked. I still went out and got a 750W PSU when one came up on sale... but

Mr Crispy

OcUK main man Gibbo just posted this on their forum, re Ampere availability;

Hi there

So a little more info for you guys.
It is still early days but on the 17th we will have stock on 3080's, volumes it is still to soon to comment but even if we get 100 or 1000 cards they will sell out quickly, the amount of searches on our webshop for the 3080 is into the thousands.

Also 3090 shall be in stock on 24th also, but again same situation as 3080, the demand is looking to be huge.

They're also limiting purchases to one per customer.

BloodBath

Samuel T. Cogley wrote: »

Disagree on the slow updates to AMD drivers for GPUs. We got three updates in August, one of them was even working properly.

I hadn't had an update in ages, realised I only had recommended updates enabled. I thought the optional stuff was beta.

Hopefully this is more stable. The reason for me ****ting on their drivers is having a 5700 for nearly a year now that still has driver issues.

Squidgy Black wrote: »

A lot of build calculators are still saying 650w is fine with the TDP for a 3080 once you're not rocking a huge TDP cpu like the 10900k that Nvidia used for testing. For example my Ryzen 5 3600, 2x8gb DDR4 sticks, 2 SSDs + NVME, HDD, 4 Fans and an AIO are showing at a max load wattage of 540w on OuterVision when you add a 3080 into the equation.

Now that's without any overclocking on the either the CPU or GPU, when you start pushing up on either the load gets closer to around 600w

That's absolute peak, you're practically never going to get near that. Most good quality PSU's can handle spikes above their rating for short periods as well.

frozenfrozen

Cordell wrote: »

The issue is with the MB/CPU/Firmware not supporting SR-IOV, not with the card itself. So it may be working, but only on server motherboards.

Ok well bios with sr-iov have been available for a lot of consumer boards since x79 and I've seen a lot of threads about sr-iov working with TRX40 so I don't think that will be an issue.

It's usually disabled on the card itself, it throws error 43 or something if it detects a virtual machine. I'm led to believe that won't be the case with 3000s which would be very nice

Squidgy Black

K.O.Kiki wrote: »

You can't trust any calculators if no one has benchmarked the cards yet.

I'm not saying to blindly trust them, but Nvidia have said the max TDP is 320w. It's not hard to then add up the rest of the TDP for your system with a calculator.

Unless Nvidia are changing their specs and the load is infact another 100w.

Cordell

The issue is with the MB/CPU/Firmware not supporting SR-IOV, not with the card itself. So it may be working, but only on server motherboards.

frozenfrozen

Cordell wrote: »

But not on today's desktop grade MBs and CPUs...

Source? it's usually disabled purposefully unless you go to their very high end cards. That announcement says it's working

cookie1977

K.O.Kiki wrote: »

You can't trust any calculators if no one has benchmarked the cards yet.

Nvidia are quoting 750W recommended with an intel processor. They've said the 3080 TGP is 320W. Most likely they are being conservative. I think even before benchmarks we can take educated guesses and be right.

K.O.Kiki

Squidgy Black wrote: »

A lot of build calculators are still saying 650w is fine with the TDP for a 3080 once you're not rocking a huge TDP cpu like the 10900k that Nvidia used for testing. For example my Ryzen 5 3600, 2x8gb DDR4 sticks, 2 SSDs + NVME, HDD, 4 Fans and an AIO are showing at a max load wattage of 540w on OuterVision when you add a 3080 into the equation.

Now that's without any overclocking on the either the CPU or GPU, when you start pushing up on either the load gets closer to around 600w

You can't trust any calculators if no one has benchmarked the cards yet.

cookie1977

jonski wrote: »

I'm guessing here, and going to wait until proper reviews are out but I'm thinking my 3 year old EVGA SuperNOVA 650 P2, 80+ PLATINUM 650W is going to be as bit under what I would need for the 3080 ?

Unless you're oveclocking I cant see why the platinum 650W wouldn't be enough. I plan on using a Corsair 650 80+ goldwith my 3080 and amd 3900x cpu. Whats your processor?

Samuel T. Cogley

grogi wrote: »

Which ones? I don't want to be guinea pig here

20.8.3 seem pretty stable. I was on 20.4.2 previously I think.

Squidgy Black

A lot of build calculators are still saying 650w is fine with the TDP for a 3080 once you're not rocking a huge TDP cpu like the 10900k that Nvidia used for testing. For example my Ryzen 5 3600, 2x8gb DDR4 sticks, 2 SSDs + NVME, HDD, 4 Fans and an AIO are showing at a max load wattage of 540w on OuterVision when you add a 3080 into the equation.

Now that's without any overclocking on the either the CPU or GPU, when you start pushing up on either the load gets closer to around 600w

Cordell

It depends on the PSU, but good ones will have the overcurrent protection saving everything else, including themselves.

But there may be another issue, pulling more power than the spec on a single PCIe cable while still being under the PSU limit, this is the really problematic one.

KillerShamrock

jonski wrote: »

I'm guessing here, and going to wait until proper reviews are out but I'm thinking my 3 year old EVGA SuperNOVA 650 P2, 80+ PLATINUM 650W is going to be as bit under what I would need for the 3080 ?

There is a lot of ah sure it be enough floating around not just here but other forums and such too. It maybe the case to be honest.
Or 650 power supplys will be the 2080ti of PSUs ...ðŸ˜‰

What would happen to a computer that has a decent 650w and the 3080 brings it over the limit (all other parts being reasonable like a 3700x or 3800x) will the PSU just crap out, explode or just force the system to shut down?

jonski

I'm guessing here, and going to wait until proper reviews are out but I'm thinking my 3 year old EVGA SuperNOVA 650 P2, 80+ PLATINUM 650W is going to be as bit under what I would need for the 3080 ?

Azza

BloodBath wrote: »

Exactly. AMD are miles behind Nvidia in the gpu market and I don't think RDNA2 is going change that.

They have a bad rep for a reason, crap drivers, not updated often enough, terrible stock coolers, not competitive in the high end for over 5 years. Nvidia have already capitalized on that and are not giving AMD an inch to catch up. They are only extending their lead.

AMD need massive investment into their gpu division, both in hardware and software if they want to compete again and change their deserved poor image in the gpu market. Even selling at little to no profit to try and regain some reasonable market share and public image.

AMD drivers being poor use to be true but has not been the case for a year or more, and there released at a rate of one a month, same as NVIDIA. But stock coolers is a fair complaint though. AMD have also proven to be rather crap at marketing.

As for not being competitive a the end high, its true at the ultra high end but I'd still classify card like the GTX 1070/1080 has high end cards, at least when looking at pricing, that's pricing I would consider to be high end GPU.

Vega was late and power hungry compared to Pascal but they where competitive in terms of performance. Vega 56 beat the 1070 and the Vega 64 though slower at first than the GTX 1080 arguably had better legs due to it handling DX12 and Vulkan better than the Pascal architecture ever did.

The gap widened last generation, AMD have no counter to the RTX 2070 Super and above (Vega 7 was just a stop gap counter) though you could argue the 5700XT compete on value against the RTX 2070 Super. Feature wise I don't think RT and DLSS have been essential features this generation of GPU's as the technology need time to be implemented into games but going forward AMD does need to have a counter to them in in Big Navi as RT and DLSS look set to become widely implemented features.

I think the main issue AMD had over the last years was money. They pumped everything into the CPU side of things and on the GPU side where more focused on the console side of things to secure a steady cash flow. They where in trouble on all fronts with their CPU and GPU's and with limited resources decided to focus on the CPU side of things to steady the ship at the expense of the GPU division. Looks to have been the right move and now that they are on more stable financial ground they may be able to be divert more resources into the GPU side of things.

grogi

Samuel T. Cogley wrote: »

Disagree on the slow updates to AMD drivers for GPUs. We got three updates in August, one of them was even working properly.

Which ones? I don't want to be guinea pig here

Samuel T. Cogley

Disagree on the slow updates to AMD drivers for GPUs. We got three updates in August, one of them was even working properly.

Cordell

Some all in one CPU/GPU/VRAM may suit the laptop and console market but for the desktop users will take away one of the most important feature, which is the ability to upgrade only what needs to be upgraded.
And it was already tried with no real success by Intel: https://www.engadget.com/2018-01-07-intel-amd-rx-vega-m.html

grogi

BloodBath wrote: »

As long as the step up is big which it will be then there's always a market for it.

I would love to see AMD attempt a high end desktop APU that maybe even bundles in some HBM graphics memory on the package but for whatever reason they aren't doing it.

I think down the line that may well be the way things go. The majority of power usage in PC's is spend transferring data. The further it has to go the less efficient it's going to be. Bundling everything onto the 1 package is probably the future but it doesn't seem too close.

We already have heat issues, the zen2 CPUs are so tight that they cannot generate any more heat. If you start bundling all together, you would have to throttle parts down to conserve the heat budget.

K.O.Kiki

Macker1 wrote: »

I was looking for a white coloured GPU for a recent white themed build. Not many options at the current 2000 series. Any thoughts on when or if white coloured 3000 series GPU's would become available.

Very niche market based on lack of options.

You'll have to import from China.

I also love their LEGO GAMER card :pac:

BloodBath

Hyzepher wrote: »

I agree with most of this and that's why I prefaced my comments with the statement that this is unlikely until the step up in GPU performance no longer matters to the general gamer.

As long as the step up is big which it will be then there's always a market for it.

I would love to see AMD attempt a high end desktop APU that maybe even bundles in some HBM graphics memory on the package but for whatever reason they aren't doing it.

I think down the line that may well be the way things go. The majority of power usage in PC's is spend transferring data. The further it has to go the less efficient it's going to be. Bundling everything onto the 1 package is probably the future but it doesn't seem too close.