guru3d Rumored GeForce 5090 Specs

[EHW Ai]

  On 06/01/2024 at 21:28, UltraMega said:

 

KInda a side note that your comment made me think about, what do you guys usually do with your old GPU when you upgrade? For a long time now I have been in the habit of selling my old GPU right after I get a new one, so the price to upgrade is only ever as high as the difference between what I can sell my old one for and what the new one costs. 

 

Because of Covid/mining, when I sold my 1080Ti and upgraded to an RX6800, I actually made a little bit of money (like $10). After that I sold my RX6800 for $400 and bought the 7900XT for $800. 

 

Do you guys do this sort of thing? 

Expand  

 

I typically give away things to those deserving, mostly because I can't take it with me.

 

Now that I could use a new to me  truck though...

 

 

 

 

something like those would do

 

Edited January 7, 2024 by iamjanco

[Snakecharmed]

  On 07/01/2024 at 02:40, Slaughtahouse said:

Thanks for sharing the additional context about the US market  I’ll stop derailing this thread

Expand  

 

Likewise with the Canadian market! I admit I don't know much about ARMs because I never shopped them before, so I'm probably wrong about the most popular ones. In addition to 5/1 and 5/5, I know there are other terms like 3/1 and also some less common ones like [5|7|10]/6 as well. The interest rates seem comparable to the Canadian 5-year, and the amortization on ARMs are typically 30 years.

 

A US fixed rate mortgage typically amortizes at the end of the advertised term, so I was slightly confused when first reading that five years is the most typical "long" fixed rate mortgage in Canada because I figured there was no way that could also be the amortization period.

 

Anyway, some other thoughts on the GPU-flipping subtopic:

 

I've come to realize I don't have a need for a spare GPU to put in another rig, which is why I try to sell off the old one as soon as I get a new one. If I ever have a dead GPU, I can run iGPU while I shop for a new one.

 

I've mentioned in another thread before that I used to dabble with eGPU setups for my laptop, but that ultimately never went anywhere. If I were to ever build a second concurrent rig rather than use a laptop or mini PC, it would be for an HTPC, which would still be small form factor and not have the space or a beefy enough PSU to run a full-sized card that used to be in my desktop.

 

I've taken to the mentality of any compute-oriented electronics I'm not actively using are a depreciating asset, so get rid of them as soon as conveniently possible. The only time that wasn't true was the convergence of the pandemic/supply chain shortages/second mining boom.

Edited January 7, 2024 by Snakecharmed

[Sir Beregond]

  On 06/01/2024 at 21:28, UltraMega said:

 

KInda a side note that your comment made me think about, what do you guys usually do with your old GPU when you upgrade? For a long time now I have been in the habit of selling my old GPU right after I get a new one, so the price to upgrade is only ever as high as the difference between what I can sell my old one for and what the new one costs. 

 

Because of Covid/mining, when I sold my 1080Ti and upgraded to an RX6800, I actually made a little bit of money (like $10). After that I sold my RX6800 for $400 and bought the 7900XT for $800. 

 

Do you guys do this sort of thing? 

Expand  

I always hang on to cards as "emergency" cards. Problem with that approach is the increasing cost of new cards and the decreasing value of held onto cards as they increasingly become obsolete and less valuable as an "emergency card".

 

I guess now that HWBot is a thing though, there will always be something for me to do with old hardware.

[ENTERPRISE]

  On 06/01/2024 at 21:28, UltraMega said:

 

KInda a side note that your comment made me think about, what do you guys usually do with your old GPU when you upgrade? For a long time now I have been in the habit of selling my old GPU right after I get a new one, so the price to upgrade is only ever as high as the difference between what I can sell my old one for and what the new one costs. 

 

Because of Covid/mining, when I sold my 1080Ti and upgraded to an RX6800, I actually made a little bit of money (like $10). After that I sold my RX6800 for $400 and bought the 7900XT for $800. 

 

Do you guys do this sort of thing? 

Expand  

 

Definitely, to be honest it is the only way I can "Justify" the upgrade to a newer generation. In order to avoid downtime I try to purchase the new GPU and get it in my system first, then will sell the older GPU to offset the cost.

 

The 3090 I have now is the single most expensive card I have ever purchased, and that was even with using the EVGA queue system before everything went nuts with Covid/Mining. I think I will be preserving my 3090 for some time to come. As much as I would enjoy a 5090, I reckon the price of it (assuming Nvidia keep on trend) will put it outside the realm of justifiable. Gone are the days of buying this "Because I can" lol.

[UltraMega]

How do you guys determine the used prices? Do you guys usually sell your GPUs on Ebay? 

 

I feel like it is absolutely essential to be effectively upgrading when you are in the high end. 

Edited January 7, 2024 by UltraMega

[ENTERPRISE]

  On 07/01/2024 at 12:52, UltraMega said:

How do you guys determine the used prices? Do you guys usually sell your GPUs on Ebay? 

 

I feel like it is absolutely essential to be effectively upgrading when you are in the high end. 

Expand  

Ebay for sure. I take a look at the competition, AKA other sellers and I choose to undercut them if I would like a quick sale....or match them if I am happy to wait.

[neurotix]

Yeah I love my 4090 for f@h and since I only game at 3440x1440p 144Hz (and mostly play Forza Horizon series, a very old MMO, and have a bunch of older games to play through) I wouldn't need the upgrade. Horizon 4 and 5 are extremely well optimized and DX12, my 1080tis in SLI I upgraded from couldn't do 144Hz whereas the 4090 does it at ease with only 70% GPU usage and runs at 45C.

 

I'm drooling over those specs if they turn out to be true but I just don't play enough modern games to justify the purchase... And *ahem* justify the purchase to my fiancee in particular

 

I will probably be hanging on to the 4090 for another 4 years and then get a 6090 or 7090 but if they are priced above $2k USD I might have to look at alternatives or go for a lower tier card.

 

That's my two cents.

Edited January 7, 2024 by neurotix

[Fluxmaven]

I look computer components as a sunk cost in entertainment. I'd get more hours of enjoyment from a 4090 than a ticket to a Taylor Swift concert for the same cost. Granted I wouldn't enjoy one of her shows if it were $20 lol.

 

Eventually I'll probably get tired of the hobby and sell everything off for cheap, but at that point I'll have had years of enjoyment from everything. 

[neurotix]

  On 07/01/2024 at 19:58, Fluxmaven said:

I look computer components as a sunk cost in entertainment. I'd get more hours of enjoyment from a 4090 than a ticket to a Taylor Swift concert for the same cost. Granted I wouldn't enjoy one of her shows if it were $20 lol.

 

Eventually I'll probably get tired of the hobby and sell everything off for cheap, but at that point I'll have had years of enjoyment from everything. 

Expand  

With everything you have you really need to get on hwbot and bench it all.

[Fluxmaven]

  On 07/01/2024 at 23:20, neurotix said:

With everything you have you really need to get on hwbot and bench it all.

Expand  

A lot of it is stuff I bought just for hwbot. Problem is I don't have time to sit around benching all day. I run the stuff in the one category of the competition that's happening at the time, then it goes on the shelf. Now that I have my stuff better organized I plan on going through and benching more. Just been kinda burnt out on computer stuff lately. 

[Slaughtahouse]

Off topic to Blackwell rumours, remember that Intel is poised to release Battlemage this year as well.

 

Which should make the mid range much more competitive against both AMD and Nvidia. 

 

https://videocardz.com/newz/intel-reaffirms-xe2-battlemage-is-coming-hardware-team-already-working-on-celestial

[bonami2]

  On 10/01/2024 at 23:57, Slaughtahouse said:

Off topic to Blackwell rumours, remember that Intel is poised to release Battlemage this year as well.

 

Which should make the mid range much more competitive against both AMD and Nvidia. 

 

https://videocardz.com/newz/intel-reaffirms-xe2-battlemage-is-coming-hardware-team-already-working-on-celestial

Expand  

 

Completly forgot about intel.

 

Calculated fast they says 60%+ more core. And i imagine it a new node.

Seem it going to be around 6800xt performance at a minimum. 

 

Not bad considering the 6800xt can beat a 4070 in some bench.

[GanjaSMK]

  On 06/01/2024 at 01:41, Slaughtahouse said:

*SNIPPED*

Expand  

 

Oh man apologies if this slides slightly off topic momentarily - BUT - holy lord it's not just the interest rates. It's clear what the writing on the wall is. I will refrain from a rabbit hole so take this only piece I'll offer with whatever you may: 

 

The biggest factor in our next worldwide recession is going to be credit debt. It won't be homes or stocks at first, it will be credit. Credit markets are in terrible, terrible shape, and very little is being discussed openly about the ramifications it will bring if defaults continue and eventually end up collapsing in large swaths. 

 

And then, back on the GPU wagon and pricing with specs - my take is awful plain - again writing on the wall. 

 

1. AI has not scaled down yet - but will by 2027 (think TinyML and more). 

2. Compute power needs a lot of energy moving post 2030 - that's not even in place yet (fission/fusion/nuclear). 

3. Battery storage is at all time high on tech - but not large enough for offline/offsite applications at large scale (also slate for post-2030 phasing). 

4. Space Race is actively ongoing - private and public (we realized humans suck at space, that's why AI / robotics). 

5. Rise and repeat. 

 

To segregate all the pricing out of this for GPU's will only follow what the market demands and unfortunately, because of the above assertions, I find it unlikely prices to ever come down any time soon. 

 

'Tis a new, rough, expensive era. 

IMHO ~ FWIW ~

[Barefooter]

Nvidia RTX 5090 could have up to 77% more memory than 4090

 

From techradar

  Quote

The Nvidia GeForce RTX 5090 has long since been a hot topic in the tech rumor mill as possibly the best graphics card in the future market, with the latest ones revealing even more information about its memory specifications.

 

According to well-known and reliable Twitter hardware leaker Kopite7kimi, the RTX 5090 will feature a 512-bit memory bus that is 33% wider than the one on Nvidia’s RTX 4090. This would allow for higher levels of memory bandwidth and increased GPU memory capabilities. It could even have 32GB of VRAM thanks to its two 16GB GDDR modules.

 

There’s also a report that the RTX 5090 will have 28Gbps GDDR7 memory modules, 33% faster than the 21Gbps memory modules in the RTX 4090. But what’s most impressive is the memory bandwidth for the alleged card, an apparent 77% boost compared to the 4090 that would give it 1,800 GB/s of memory bandwidth. This checks out with a previous rumor stating that the card would be up to 70% faster than the 4090.

 

 

If this RTX 5090 turns out to be true, the specs would be absolutely mind-blowing and would nearly justify what would certainly be a massive price hike compared to current-gen Nvidia cards.

The RTX 5090 could be in trouble

What’s interesting is that other memory producers have been promoting 32Gbps GDDR7 modules, meaning that if these rumors are true then Nvidia is purposefully slowing down the graphics cards to reduce costs and increase yields. But thanks to the Super series refresh of Nvidia cards, we could possibly see an updated version without these restrictions.

 

However, there seems to be one major caveat, which is the release date. It seems that Nvidia’s 5090 will be released in 2025, meaning that the rumored Nvidia RTX 4090 Super and RTX Titan would have to keep Team Green’s lead until then. A prospect made much more difficult by AMD.

Team Red is prepping its own comeback with the mid-range RDNA 4 graphics cards, which are shaping up to be Nvidia RTX 4080 and 4060 Ti killers. Of course, it’s possible for it to also be released in 2025, but if it comes out before Team Green’s RTX 5000-series then Team Red will still have a major advantage.

We’ll have to wait and see what happens, but regardless we’re sure to have some great competition in the graphics cards market.

Expand  

 

[Slaughtahouse]

  On 12/03/2024 at 17:42, Barefooter said:

Nvidia RTX 5090 could have up to 77% more memory than 4090

 

From techradar

 

Expand  

Game Developers Conference (GDC) and Nvidia's GPU Technology Conference (GTC) is also next week.

 

We won't hear about consumer class GPUs (SKUs, hardware, pricing etc.) but...I do expect some form of update on Blackwell architecture. Whether it's the enterprise counterpart and/or the technology we can expect to come down the pipe, they will want to preach to devs next week (and kick start their new FY... xD).

[Sir Beregond]

Unless they are going for a Titan branded product again that really is top of the line in both spec and price, I seriously doubt you'll see anything more than a 384-bit bus still. Density on GDDR7 modules is supposedly still going to be the same as what GDDR6X is now (16Gb which is 2GB per module) starting out, so unless they clamshell the memory modules to get 48GB in 3090 style, then it's probably another 24GB card.

 

There will still be an increase in bandwidth at least going from G6X to G7 as the 4090 is 21Gbps and the G7 is rumored to be 28-32Gbps.

 

GeForce 200-series (circa 2008-2009) was the last time they did any consumer cards with larger than 384-bit for traditional GDDRx based cards.

 

Color me extremely skeptical of 512-bit rumors.

Edited March 14, 2024 by Sir Beregond

[Barefooter]

Next-gen GPUs likely arriving in late 2024 with GDDR7 memory — Samsung and SK hynix chips showed chips at GTC

From Tom's Hardware

Next-gen GPUs likely arriving in late 2024 with GDDR7 memory

 

  Quote

Samsung and SK hynix were showing off their upcoming GDDR7 memory solutions at GTC 2024. We spoke with the companies, as well as Micron and others, about GDDR7 and when we can expect to see it in the best graphics cards. It could arrive sooner than expected, based on some of those discussions, and that has some interesting implications for the next generation Nvidia Blackwell and AMD RDNA 4 GPUs.

Micron didn't have any chips on display, but representatives said their GDDR7 solutions should be available for use before the end of the year — and the finalized JEDEC GDDR7 standard should pave the way. The chips shown by SK hynix and Samsung meanwhile would have been produced prior to the official standard, though that may not matter in the long run.

What's interesting about the information shown by SK hynix and Samsung is that both were showing off 16Gb (2GB) devices. We asked the companies about memory capacities and were told that 16Gb chips are in production and could show up in shipping products by the end of the year. 24Gb (3GB) chips on the other hand aren't going to be in the initial wave, and will very likely show up in 2025.

 

That could make sense, as there were also rumblings at GTC 2024 that consumer Blackwell GPUs could begin shipping before the end of the year. Whether AMD RDNA 4 will also ship this year was less clear, but of course this was an Nvidia conference and AMD wasn't in attendance.

Historically, seeing an "RTX 5090" Blackwell card in the October ~ December timeframe would make perfect sense. Nvidia has been pretty good about having a two-year cadence for new GPU architectures, going back a decade. The Maxwell GTX 900-series launched in September 2014, Pascal GTX 10-series arrived in May 2016, and Turing RTX 20-series came out in September 2018. Then we got Ampere and RTX 30-series in September 2020 — just in time for a cryptocurrency surge to ruin everything for the next two years — and finally Ada Lovelace and RTX 40-series launched in October–November 2022.

There have been questions about whether Blackwell consumer GPUs would arrive in 2024, and the relatively late launch of the RTX 40-series Super update at least hinted at a potential push of the next generation. But a few people that I spoke with at GTC basically said the Super cards were "late" and that Blackwell consumer parts likely wouldn't be affected. So... RTX 50-series coming in late 2024? That seems increasingly likely after what we saw at GTC, including the massive Blackwell B200 reveal.

Current Blackwell GPU rumors suggest the top consumer model GB202 will feature a 512-bit memory interface, while the step down GB203 will only have a 256-bit interface. After seeing GB200 with its dual-die solution that links via NV-HBI (Nvidia High Bandwidth Interface), we can't help but think that's at least potentially on the table for consumer chips as well. Imagine if GB202 is simply two GB203 chips — which would then explain the doubling in memory bus width. Otherwise, another 384-bit interface would have seemingly made more sense.

Beyond that, however, there are other implications. 16Gb GDDR7 chips are basically ready and will be shipping in volume later this year... just in time for that supposed 512-bit memory interface. Sixteen chips working together would yield 32GB of VRAM, and that seems like a perfectly sensible configuration for a future "RTX 5090" — or whatever Nvidia wants to call it. But the step down to a 256-bit interface for "RTX 5080" is a bit less impressive.

Using the same 16Gb GDDR7 chips would give us 16GB of VRAM on the second tier GPUs. That's certainly "enough" for most current workloads, outside of AI, but AMD has been shipping 16GB cards for four years now. The alternative would be to wait for 24Gb GDDR7 devices to become available, which would mean 24GB on a 256-bit interface. We'd prefer that over yet another 16GB card, but we'll have to see how things play out.

Expand  

 

[Sir Beregond]

  On 26/03/2024 at 14:48, Barefooter said:

Next-gen GPUs likely arriving in late 2024 with GDDR7 memory — Samsung and SK hynix chips showed chips at GTC

From Tom's Hardware

Next-gen GPUs likely arriving in late 2024 with GDDR7 memory

 

 

Expand  

My guess is 5090 is either another 24GB card or 48GB if they decide to clamshell 3090 style. And the 5080 if it's 256-bit again, is another 16GB card then, unless it comes out when the 24Gb chips arrive in which case it becomes possible for 24GB as well. I could see 48GB, then 24GB. And then if the 70 and 60 cards also use higher density memory into 2025, enables options for some odd sizes like 18GB for 192-bit cards and 12GB for 128-bit.

 

Or they could just go lame and pull another 40-series lineup VRAM wise. Who knows.

Edited March 26, 2024 by Sir Beregond

[Barefooter]

  On 26/03/2024 at 15:18, Sir Beregond said:

Snip

Or they could just go lame and pull another 40-series lineup VRAM wise. Who knows.

Expand  

That would not surprise me in the least!

Edited March 26, 2024 by Barefooter

[ozlay]

  On 26/03/2024 at 15:18, Sir Beregond said:

My guess is 5090 is either another 24GB card or 48GB if they decide to clamshell 3090 style. And the 5080 if it's 256-bit again, is another 16GB card then, unless it comes out when the 24Gb chips arrive in which case it becomes possible for 24GB as well. I could see 48GB, then 24GB. And then if the 70 and 60 cards also use higher density memory into 2025, enables options for some odd sizes like 18GB for 192-bit cards and 12GB for 128-bit.

 

Or they could just go lame and pull another 40-series lineup VRAM wise. Who knows.

Expand  

36GB

[Sir Beregond]

  On 26/03/2024 at 18:59, ozlay said:

36GB

Expand  

That would be nice, but sounded like no 24Gb chips till next year, so if the 5090 launches this year, its stuck with 2GB modules so 24GB or 48GB if a 384-bit card.

 

I personally don't buy the 512-bit rumors.

[UltraMega]

The short:

 

 

And the long:

 

 

 

I'd love to see the commentary on this. 

[The Pook]

 

  On 26/03/2024 at 22:08, Sir Beregond said:

That would be nice, but sounded like no 24Gb chips till next year, so if the 5090 launches this year, its stuck with 2GB modules so 24GB or 48GB if a 384-bit card.

 

I personally don't buy the 512-bit rumors.

Expand  

 

me neither 

 

half expecting to get another generation with the same VRAM limits on the lower tier (<5090) cards honestly, seems like an NVIDIA move  

[Sir Beregond]

  On 27/03/2024 at 20:45, The Pook said:

 

 

me neither 

 

half expecting to get another generation with the same VRAM limits on the lower tier (<5090) cards honestly, seems like an NVIDIA move  

Expand  

I agree. The only difference is I don't think they'll try to launch a 5060 8GB. I think they'll try to hold out for the higher density memory chips into 2025 before launching those cards. If they can get 3GB modules, then even a 128-bit card can have 12GB.

[Section31]

  On 27/03/2024 at 21:42, Sir Beregond said:

I agree. The only difference is I don't think they'll try to launch a 5060 8GB. I think they'll try to hold out for the higher density memory chips into 2025 before launching those cards. If they can get 3GB modules, then even a 128-bit card can have 12GB.

Expand  

 

Well the 5090 will mark my return to the PC building side (from my vacation exodus on the custom keeb side).