So some systems have gpus that put out more heat.
Why are load temps better for gpu's than on cpus?
So some systems have gpus that put out more heat.
Why are load temps better for gpu's than on cpus?
Last edited by homefry; 05-19-2011 at 09:32 AM.
Thats a pretty ridiculous generalization, if anything in a standard air cooled system the GPUs get considerably hotter under heavy load.
As your talking in the water cooling forum i presume you mean water cooled CPU vs GPU.
The load temp is better because the GPU chip is bigger as is the waterblock cooling the GPU, so you have more surface contact area for the GPU. ie more heat is removed over a given time due to more time water is in contact with the block.
My CPU runs at 70c compared with 50c for 4 GPUs running in series.
Last edited by Phatboy69; 05-18-2011 at 10:03 PM.
I am Intel of Borg. Resistance is futile. You will be assimilated. Borg Homeworld - Blog
i7 3930k @ 5Ghz AC Kryos Silver - Asus Rampage IV Extreme - 4 way SLI 3GB GTX 580-UD @ 1000/2200 - 16GB Corsair GT DDR3-2000 RAM - 4 x GTX 360 rads w/ 24 x CM push-pull fans & shrouds - Aquaero 5 XT controller, 6 PA2 Ultras - 2 Flow meters - 2 x Enermax 1500W PSU - MM Extended Ascension Horizon XL-ATX Case - 2 x Koolance 452x2 Res and 4 x D5 pumps - 3 x 24" Acer H243H Surround LCD
1/ Greater heat load (sometimes)
2/ Larger die surface (i.e. Smaller Heat Flux - i.e. easier to cool)
3/ Probe location (high gradients of temperature throughout the die surface)
4/ IHS vs no IHS (on AMD at least..)
#2 + #3 make CPU and GPU not really comparable temperature wise
Is it just me or is the question in the title the inverse of the one in the post?
subtec: True
EK Water Blocks R&D
ekwb.com
Hobby: Rotary SS bench unit (LPG/R-507) | -75°C Auto-cascade (R-600/R-290/R-744) | Plate HX water chiller (1kW cooling power) | My 2nd cascade MP14FB R-507/R-1150 | Mini cascade R-507/R-1150 | GPU single-stage | Mach I restoration mod -> -38°C @ 230W
Milestones: 7.12GHz P4 | 6.84GHz K10.5 | 6.45GHz Ci5 | 3.31GHz K7
3DMark2001SE -> Favourite Game!
i7 920 D0 @ 4.01ghz 1.25v
Rampage III Extreme
24gb Corsair Vengeance @ 1600mhz 8-8-8-24-1T
eVGA GTX 1070 Founders Edition
Silverstone 1000w
Corsair 800D
Liquid Cooled - CPU/MB - EKWB Supreme HF/Full Cover MBWB, Iandh 225 res, D5 w/XTOP rev. 2, Feser x-changer 360, Bitspower fittings
yeah, that was a ridiculous mistake. i've been studying too long.
You would think overtime, all your components would stabilize toward the same temp within at least 10C and not a crazy high 20C delta.
Even GPU only blocks the same size as cpu only blocks get better temps, so it doesn't have to do with the size of gpu full cover blocks.
Last edited by homefry; 05-19-2011 at 12:14 PM.
1. Some systems with a 5970 and 920 have lower temps on the gpu even though the gpu puts out more heat.
3 This is a non issue because most probes are within the hottest areas.
4. IHS should have a minimal impact. Surely not as much as 20C delta we see in systems.
You cannot compare two different sensors, it's like comparing apples to oranges.
EK Water Blocks R&D
ekwb.com
Hobby: Rotary SS bench unit (LPG/R-507) | -75°C Auto-cascade (R-600/R-290/R-744) | Plate HX water chiller (1kW cooling power) | My 2nd cascade MP14FB R-507/R-1150 | Mini cascade R-507/R-1150 | GPU single-stage | Mach I restoration mod -> -38°C @ 230W
Milestones: 7.12GHz P4 | 6.84GHz K10.5 | 6.45GHz Ci5 | 3.31GHz K7
3DMark2001SE -> Favourite Game!
by size of die... referring to power density of cpus are higher. 150W consumed in size of 1 fingerprint it going to have a higher temp gradient than 200W consumed in size of 2 fingerprints. Every time the die shrinks, power density rises, so architectural improvements and smaller v must counterbalance rise.
Bookmarks