So, who's getting a 5XXX series card?

[Woomack]

Somehow, many more half-height PC cases were released in recent years, and I wonder why, when people can't find proper components for them. The same with some cases that support Flex-ATX PSUs or weird non-standard PSUs. Those are mainly from lesser-known Chinese brands, but in Asia, they sell quite well.

There were single half-height graphics cards with a power connector, but I don't remember what they were. They were still lower GPUs at ~100W. When I was testing some of the new graphics cards, I noticed they barely use the PCIe slot to deliver power. Almost everything they take from the power connector on top of the card. Half-height could still have it on the side, but there will still be a problem with cooling.

For some reason, new 300mm, 2-slot, and not too tall cards are now called SFF versions. I saw many negative comments on the web about it, as it's hard to call it SFF, but all leading brands and Nvidia are pushing this idea. In the end, you make an ITX PC that is 30% smaller than a typical large mid-tower. This is why Corsair 2000D or the latest Fractal ITX cases don't sell. They are ITX, but take almost as much space as a typical large tower.

 

[mackerel]

Somehow, many more half-height PC cases were released in recent years, and I wonder why, when people can't find proper components for them. The same with some cases that support Flex-ATX PSUs or weird non-standard PSUs. Those are mainly from lesser-known Chinese brands, but in Asia, they sell quite well.

I nearly bought a Jonsbo N4 for NAS but combined with a new small PSU for it. With the cost, I aborted and just went for another Node 804. If I followed through with the N4, if I wanted a GPU I'd need a low profile one so that was another strike against it. I guess it can make more sense if you're a system builder and want to sell smaller boxes. In many places in the East, land prices are high so space is a premium. Large towers aren't so desired there.

There were single half-height graphics cards with a power connector, but I don't remember what they were. They were still lower GPUs at ~100W. When I was testing some of the new graphics cards, I noticed they barely use the PCIe slot to deliver power. Almost everything they take from the power connector on top of the card. Half-height could still have it on the side, but there will still be a problem with cooling.

While PCIe slot is nominally 75W there is some complication in that you only get 66W from 12V, and the remainder from other power rails. So extracting and balancing that might mean they'd just rather add a power connector.

Of the PSUs not needing a power connector:
750 Ti: 60W
1030: 30W
1030 DDR4: 20W
1050 (Ti): 75W
1650: 75W
3060 6GB: 70W

There were some pro ones above those, but I'm not listing them as they're unaffordable for mortals. I still think we're due a new consumer tier model in that area. We didn't get an Ada consumer one, so maybe Blackwell? The increase in power efficiency should allow a perf increase over Ampere at the same power limit.

For some reason, new 300mm, 2-slot, and not too tall cards are now called SFF versions. I saw many negative comments on the web about it, as it's hard to call it SFF, but all leading brands and Nvidia are pushing this idea. In the end, you make an ITX PC that is 30% smaller than a typical large mid-tower. This is why Corsair 2000D or the latest Fractal ITX cases don't sell. They are ITX, but take almost as much space as a typical large tower.

Wasn't that the Nvidia definition? I guess the thing about SFF is how small is small? I liked the SG13 case and had two in the past. That would take a 2 slot GPU (preferably blower) as well as ATX PSU. If you want to go smaller, that does limit things further, so maybe there needs to be SFF GPU classes, not just the one.

 

[Janus67]

TL;DW -- woof, don't even attempt to recommend the 8GB card. It has significantly worse performance than the 16GB model.

It is okay at 1080p medium and low settings, but who is buying a $400 card to target 1080p/medium settings in 2025? It's too expensive for 1080p and too underpowered to run at 1440p/4k. What a step down from 2 generations ago there was at least a 12GB 3060ti.

 

[mackerel]

NVIDIA GeForce 576.02 driver said to be affected by GPU temperature sensor bug - VideoCardz.com

Users report temperature sensor bug in NVIDIA 576.02 drivers There’s apparently a new bug in NVIDIA graphics drivers affecting temperature readings. The GeForce 576.02 driver update, released to address black‑screen and stability problems, may still cause issues. Some users report that GPU...

videocardz.com

Claim that the current 576.02 driver might have a temperature reporting bug.

 

[EarthDog]

Can't say I noticed this..............but didn't pay attention either. I'll check it out tonight tho....

 

[mackerel]

Reading it more, it seems to possibly happen when you resume from sleep, at least according to the Nvidia bug tracking description. I see a comment it might also happen from hibernation, and that not all software might be affected.

I'm now wondering, if it does happen to someone, is it a simple mis-reading that doesn't change anything, or does it affect the actual operation of the GPU? Comments that apparently Nvidia's own software can still monitor correctly so it might just be some API thing for 3rd party tools.

 

[EarthDog]

I realized I don't sleep either so... perhaps I'm not a good test subject. ANytime I've looked at temps since I got these drivers (a couple of days ago), they moved.

 

[mackerel]

NVIDIA releases GeForce Hot Fix driver solving GPU temperature reporting issue - VideoCardz.com

NVIDIA has fixed the GPU temperature sensor problem on RTX 50 series Reports suggesting the issue is more widespread than indicated in NVIDIA’s known driver bug lists may have pressured the company to release a hotfix driver addressing the problem quicker. Earlier today, we reported that the...

videocardz.com

There's a hotfix driver that should resolve that, and some other things.

 

[EarthDog]

Sigh. So sloppy this go around, NV...

 

[SF101]

I nearly bought a Jonsbo N4 for NAS but combined with a new small PSU for it. With the cost, I aborted and just went for another Node 804. If I followed through with the N4, if I wanted a GPU I'd need a low profile one so that was another strike against it. I guess it can make more sense if you're a system builder and want to sell smaller boxes. In many places in the East, land prices are high so space is a premium. Large towers aren't so desired there.


While PCIe slot is nominally 75W there is some complication in that you only get 66W from 12V, and the remainder from other power rails. So extracting and balancing that might mean they'd just rather add a power connector.

Of the PSUs not needing a power connector:
750 Ti: 60W
1030: 30W
1030 DDR4: 20W
1050 (Ti): 75W
1650: 75W
3060 6GB: 70W

There were some pro ones above those, but I'm not listing them as they're unaffordable for mortals. I still think we're due a new consumer tier model in that area. We didn't get an Ada consumer one, so maybe Blackwell? The increase in power efficiency should allow a perf increase over Ampere at the same power limit.


Wasn't that the Nvidia definition? I guess the thing about SFF is how small is small? I liked the SG13 case and had two in the past. That would take a 2 slot GPU (preferably blower) as well as ATX PSU. If you want to go smaller, that does limit things further, so maybe there needs to be SFF GPU classes, not just the one.

Somehow, many more half-height PC cases were released in recent years, and I wonder why, when people can't find proper components for them. The same with some cases that support Flex-ATX PSUs or weird non-standard PSUs. Those are mainly from lesser-known Chinese brands, but in Asia, they sell quite well.

There were single half-height graphics cards with a power connector, but I don't remember what they were. They were still lower GPUs at ~100W. When I was testing some of the new graphics cards, I noticed they barely use the PCIe slot to deliver power. Almost everything they take from the power connector on top of the card. Half-height could still have it on the side, but there will still be a problem with cooling.

For some reason, new 300mm, 2-slot, and not too tall cards are now called SFF versions. I saw many negative comments on the web about it, as it's hard to call it SFF, but all leading brands and Nvidia are pushing this idea. In the end, you make an ITX PC that is 30% smaller than a typical large mid-tower. This is why Corsair 2000D or the latest Fractal ITX cases don't sell. They are ITX, but take almost as much space as a typical large tower.

Yea To be fair it doesnt need to be a full half height card in most cases it just needs to be level with the rear slot IO, Most smaller form factor cases for example will fit a 290X tight to the cover but if the card is taller than the rear IO your screwed.

I have ran into over a dozen situations where this would be nice to have but most of the SFF cases now just use risers and move the GPU orientation to acomodate BIG gpu's.

My point was if the 5060/TI really wanted a market that justified its existance when facing against much cheaper alternatives and its own 5070 line a SFF or half hight /low profile at the very least would really give it a niche leg to stand on since its lower power etc.

And people looking for a HTPC might find it apealing.

Kind of like how the A310-380s were basically trash but work amazingly in a low price low power AV1 Encoding/Transcoding scenerios.

 

[Railgun]

Assuming the internal space overall is sufficient to hold the card itself, it should be somewhat trivial to fab the correct I/O bracket as ultimately, that's what matters with full/half height anyway.

 

[Kenrou]

"Parts of NVIDIA GeForce RTX 50 Series GPU PCB Reach Over 100°C: Report"

"Igor's Lab has run independent testing and thermal analysis of NVIDIA's latest GeForce RTX 50-series graphics cards, including the add-in card partner design RTX 5080, 5070 Ti, and 5060 Ti, which are now attracting attention for surprising thermal "hotspots" on the back of their PCBs. These hotspots are just the areas on PCB that get hot under load, and not the "Hot Spot" sensor NVIDIA removed with RTX 50 series. Infrared tests have shown temperatures climbing above 100°C in the power delivery region, even though the GPU die stays below 80°C. This isn't a problem with the silicon but with concentrated heating in clusters of thin copper planes and via arrays. Card makers like Palit, PNY, and MSI have all seen the same issue since they closely follow NVIDIA's reference PCB layout and use similar cooler mounting. A big part of the trouble comes down to how PCB designers and cooler engineers work separately.

NVIDIA's Thermal Design Guide gives AIC partners detailed power-loss budgets, listing worst-case dissipation for the GPU, memory, NVVDD and FBVDDQ rails, inductors, MOSFETs, and other components, and it recommends ideal thermal interface materials and mounting pressures. The guide assumes that even heat is spreading and that there is perfect airflow in a wind tunnel, but actual consumer PCs don't match those conditions. Multi-layer PCBs force high currents through 35 to 70 µm copper layers, which join at tight via clusters under the VRMs. Without dedicated thermal bridges or reinforced vias, these areas become bottlenecks where heat builds up, and the standard backplate plus heat-pipe layout can't pull it away fast enough.

This hotspot issue isn't limited to the newest Blackwell GPUs. Even the previous-generation GeForce RTX 4090, which was engineered for up to 600 W of heat dissipation with multiple vapor chambers and a three-slot cooler, showed a similar pattern. Thermographic snapshots of prototype Ada architecture boards revealed rear-side VRM zones reaching the mid-70s Celsius while the GPU die sat in the low-60s. Internal versions of NVIDIA's guide were redacted to protect internal details, and no special note was made to reinforce the backplate in that spot. As a result, partners assumed the standard backplate contact area was sufficient and didn't add any extra cooling measures. Fortunately, simple tweaks can make a big difference. In Igor's Lab tests, placing a thin thermal pad or a small amount of conductive putty between the hotspot area and the backplate dropped peak VRM temperatures by 8 to 12 °C under the same load."

Parts of NVIDIA GeForce RTX 50 Series GPU PCB Reach Over 100°C: Report

Igor's Lab has run independent testing and thermal analysis of NVIDIA's latest GeForce RTX 50-series graphics cards, including the add-in card partner design RTX 5080, 5070 Ti, and 5060 Ti, which are now attracting attention for surprising thermal "hotspots" on the back of their PCBs. These...

www.techpowerup.com

 

[EarthDog]

Newsflash, gpus get hot???

In other news, water is still wet.

 

[mackerel]

Isn't the temp spec for VRMs generally way higher than high clock logic based silicon anyway?

 

[Woomack]

120°C or something, I don't remember exactly

 

[Brando]

My Hotspot says 250c in GPUz. should I worry?

 

[Brando]

Sorry I meant 255c in HWmonitor

 

[EarthDog]

Sorry I laughed, I thought you were being sarcastic...

That's an anomalous reading.

Known problem with 576.02. 576.15 is the hotfix. Try that.

 

[Brando]

Sorry I laughed, I thought you were being sarcastic...

That's an anomalous reading.

Known problem with 576.02. 576.15 is the hotfix. Try that.

It's all good man I know it's not accurate I was just mildly amused when I saw it because it fits the whole 50 series narrative.

 

[Woomack]

My ASRock Z890 OCF throws 250°C+ all over the VRM and some other sensors table, and also 200V+. Some values appear randomly, and some just go up to 250°C/200V+ when I start any test. The same is sometimes on various other components, like graphics cards. It would be good to know the real temperature or voltage, but you can't really do anything about it. Maybe some single sensors are faulty, but if it works most of the time, then probably it's just a random bad reading.