• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

5090 AIB Clocks

X

x4it3n

Joined
Oct 19, 2022
Messages
473 (0.47/day)
Location
Los Angeles, CA
System Specs
Processor AMD Ryzen 7 9800X3D (+PBO 5.4GHz)
Motherboard MSI MPG X870E Carbon Wifi
Cooling ARCTIC Liquid Freezer II 280 A-RGB
Memory 2x32GB (64GB) G.Skill Trident Z Royal @ 6200MHz 1:1 (30-38-38-30)
Video Card(s) MSI GeForce RTX 4090 SUPRIM Liquid X
Storage Crucial T705 4TB (PCIe 5.0) w/ Heatsink + Samsung 990 PRO 2TB (PCIe 4.0) w/ Heatsink
Display(s) AORUS FO32U2P 4K QD-OLED 240Hz (DP 2.1 UHBR20 80Gbps)
Case CoolerMaster H500M (Mesh)
Audio Device(s) AKG N90Q w/ AudioQuest DragonFly Red (USB DAC)
Power Supply Seasonic Prime TX-1600 Noctua Edition (1600W 80Plus Titanium) ATX 3.1 & PCIe 5.1
Mouse Logitech G PRO X SUPERLIGHT
Keyboard Razer BlackWidow V3 Pro
Software Windows 10 64-bit
5090 SUPRIM.png


We're finally seeing AIBs Boost Clocks, it's about 6-7% higher than the FE and the TDP was raised to 600W (+4.3%).
I'm curious to know what will be the Max TDP... Probably 650W since there's only 1x 16-pin connector ? :wtf:
 
The maximum wattage for a single stick is 600 watts. So unless the card has two. 600 watt is the max consumption allowed.
 
Tomgang said:
The maximum wattage for a single stick is 600 watts. So unless the card has two. 600 watt is the max consumption allowed.
Click to expand...

Wouldn't it be able to get up to 75 from the pcie slot?
 
oxrufiioxo said:
Wouldn't it be able to get up to 75 from the pcie slot?
Click to expand...
PCIe slot is specced to provide 66W on 12V (rest is 3.3V) but yes.
I was pretty surprised looking at TPU review and seeing 5090 averaging to the same 2700MHz-ish average clocks as 4090. By spec 5090 should be at a bit lower clock speeds.
 
oxrufiioxo said:
Wouldn't it be able to get up to 75 from the pcie slot?
Click to expand...
Technically yes. But i know rtx 4090 dosent pull wattage from the pci connector and the FE rtx 5090 max allowed wattage consumption seems to be 600 when power target is raised to it maximum. So it might be the same for 5090 that it dosent pull power from pci slot.

But im not sure. For that we need some one who knows more about how 5090 is constructed on that part. If it is the same as 4090. Then 600 watt is the limit that can be pulled from 1 port
 
The FE is already pulling near the power limit. I doubt aib cards will be much faster at stock.

Unless they can run at lower voltage than the FE.

Possibly they'll be some headroom if you get lucky in the silicon lottery and can manually use lower voltage with higher clocks.
 
mb194dc said:
I doubt aib cards will be much faster at stock.
Click to expand...

Yeah, you will prob be lucky to get 10%. Whether you call that a decent clock bump is debatable.

W1z got 6% without much effort and no beefy HSF or custom PCB. "Overclocking the RTX 5090 worked fairly well, and we gained +6% in real-life performance"
 
The 5090 seems power constrained even though it's already 575W lol. It would be nice to see AIB cards with a 650W or 660W TDP ! But I wonder why no AIB has 2x 16-pin... Nvidia must prevent them to do so for whatever reason.
 
x4it3n said:
The 5090 seems power constrained even though it's already 575W lol. It would be nice to see AIB cards with a 650W or 660W TDP ! But I wonder why no AIB has 2x 16-pin... Nvidia must prevent them to do so for whatever reason.
Click to expand...
I would not be so sure:
www.computerbase.de

Nvidia GeForce RTX 5090 im Test: DLSS 4 MFG trifft 575 Watt: Leistungsaufnahme und Effizienz

Nvidia GeForce RTX 5090 im Test: Leistungsaufnahme und Effizienz / Leistungsaufnahme Desktop und YouTube
www.computerbase.de www.computerbase.de
 
x4it3n said:
I still think we could get better results with more than 600W ! 2x 16-pin would have been better and safer imo.
Click to expand...
I think Nvidia stepped in and limited the power socket to one. We are looking at 665W altogether, but this card could easily do 800W with waterblock, and >1000W for LN2.
 
CornyOC said:
I think Nvidia stepped in and limited the power socket to one. We are looking at 665W altogether, but this card could easily do 800W with waterblock, and >1000W for LN2.
Click to expand...
Oh yeah definitely. Idk why Nvidia are putting so many restrictions nowadays. I remember when EVGA released the GTX 780 6GB (instead of 3GB) and would allow some crazy Overclocks on their GPUs...
But Nvidia just wanna be like Apple, everybody has the same hardware, same specs, everything is locked, and an All-Nvidia PC (we're going to see Nvidia CPUs in the next few years so yeah).
 
Last edited:
Despite the common idea that Nvidia does whatever they want it is not exactly the case. 600W is already a lot - techtubers and reviewers have very loudly said so. Nvidia not letting people send even crazier amounts of power into the thing kind of makes sense. Second connector would legitimately enable pumping 1200W to the card while one 16-pin puts a pretty hard limit on power consumption. Yes, standards and limits can be ignored but an AIB doing that is opening up a big liability for itself.

Todays small overclocks are a simple problem - while both CPUs and GPUs used to be SKUd at a pretty good point on the efficiency curve, that is no longer the case today. The stock settings today are pushed pretty high up the curve, I would say often more than average overclocks used to be back in the day. Also, dynamic clock speeds. Manufacturers are wringing out every last ounce of performance from a chip these days. Sure you could overclock say 5090 more when you can push 1200W into it but outside academic interest there is no value in there any more.
 
londiste said:
Despite the common idea that Nvidia does whatever they want it is not exactly the case. 600W is already a lot - techtubers and reviewers have very loudly said so. Nvidia not letting people send even crazier amounts of power into the thing kind of makes sense. Second connector would legitimately enable pumping 1200W to the card while one 16-pin puts a pretty hard limit on power consumption. Yes, standards and limits can be ignored but an AIB doing that is opening up a big liability for itself.

Todays small overclocks are a simple problem - while both CPUs and GPUs used to be SKUd at a pretty good point on the efficiency curve, that is no longer the case today. The stock settings today are pushed pretty high up the curve, I would say often more than average overclocks used to be back in the day. Also, dynamic clock speeds. Manufacturers are wringing out every last ounce of performance from a chip these days. Sure you could overclock say 5090 more when you can push 1200W into it but outside academic interest there is no value in there any more.
Click to expand...
I see what you're saying but trying to get every last drop of performance you can get is also very fun! :D I have a 4090 that I overclock pretty regularly for games being very hard to run like Path Tracing games where every fps matters but yeah, I only get around 10-13% more performance with my 530W BIOS (even though most games do not even reach 450W even when OC).
 
Apparently it's $2550 but I'm not 100% sure.
Either way in real world performance you won't be able to notice any difference! And with a +4% max power target on all cards, Overclocking is pretty much dead...
 
Last edited:
x4it3n said:
Apparently it's $2550 but I'm not 100% sure.
Either way in real world performance you won't be able to notice any difference! And with a +4% max power target on all cards, Overclocking is pretty much dead...
Click to expand...
That seems to be the case, top tier AIBs will struggle to justify their premium pricing.
 
CornyOC said:
That seems to be the case, top tier AIBs will struggle to justify their premium pricing.
Click to expand...
Yeah the only thing we can get this time around is better cooling but unless you get an AIO all other GPUs are MASSIVE ! It's really a turnoff for me.
Also paying $500+ for better cooling is a lot with this generation...
 
You must log in or register to reply here.
Back
Top