If you are just gaming, stick with the 12GB 3000 series 3080 and 3080 Ti. But if you need the 24GB for creative or scientific work or AI this is an insane deal, and you probably aren't overclocking, so the hot memory on the back of the card isn't a big deal.
You have no direct experience. These 3090 FTW3 cards a have 1 GB 21 Gb/s chips that are underclocked to 19.5 Gb/s from the factory. They have a big fat cooler on the front chips and a big fat backplate to the rear chips with good thermal pads. With a typical +1200 memory OC at 300w the memory junction temp reports at about 90-92C in HWiNFO with typical ambient temps and airflow. This OC and temps is consistent across multiple cards.
Furthermore, for all the alarming talk about vram junction temps, there have been no reported cases of memory failure or degradation. I got a 3080 Ti that came out of the box as an already poor memory OCer, it has not degraded after nearly a year. Heck, HIVE OS didn't even report vram junction temp until 2 months ago.
These cards have 99 problems, but memory temps is not one of them.
I do have direct experience, probably a little more than the average person. I had a 3080 Ti and have a 3090.
It all depends on the card, but my 3080 Ti FTW3 would be in the upper 70's or lower 80's depending on ambient temp. My 3090 would sometimes hit 90c+ due to just having a backplate on the memory connected with pads, and nothing else.
I never said failure, but degradation is a real thing on some 3000 series cards. Simple internet search (pick your search provider) of 3090 VRAM degradation will point you in the correct direction. The FE cards are most at fault hitting over 105c in general regardless of which FE card, but AIB cards can be tricky too. Even across identical cards, temps can fluctuate.
And as an additional note, I replaced the rear pads with a copper plate (for direct contact to the backplate) just for testing sakes, and that dropped my memory junction temps by 6c. This proves the original pads aren't the greatest.
I also added a heatsink on to the backplate afterwards and that dropped it another 10c. After all said and done, was getting in the low 70's for memory junction temp. Did I need to do all of this? Probably not, but I want my card to last a long time. I've attached pics of my mods. 20c+ memory drop when all was said and done.
Where do you think RMA'd cards go for pissed off miners that couldn't sustain their mining memory OC because they degraded their chips? Granted I don't know the percentage of this, but it's definitely not 0 from reading through various forums and posts elsewhere.
All I am saying is for people who buy this B-Stock card, or any B-Stock card, it doesn't hurt to test even a mild OC on the Memory.
Everyone makes mistakes in assuming, so I won't trash talk you because of it.
I do have direct experience, probably a little more than the average person. I had a 3080 Ti and have a 3090.
It all depends on the card, but my 3080 Ti FTW3 would be in the upper 70's or lower 80's depending on ambient temp. My 3090 would sometimes hit 90c+ due to just having a backplate on the memory connected with pads, and nothing else.
I never said failure, but degradation is a real thing on some 3000 series cards. Simple internet search (pick your search provider) of 3090 VRAM degradation will point you in the correct direction. The FE cards are most at fault hitting over 105c in general regardless of which FE card, but AIB cards can be tricky too. Even across identical cards, temps can fluctuate.
And as an additional note, I replaced the rear pads with a copper plate (for direct contact to the backplate) just for testing sakes, and that dropped my memory junction temps by 6c. This proves the original pads aren't the greatest.
I also added a heatsink on to the backplate afterwards and that dropped it another 10c. After all said and done, was getting in the low 70's for memory junction temp. Did I need to do all of this? Probably not, but I want my card to last a long time. I've attached pics of my mods. 20c+ memory drop when all was said and done.
Where do you think RMA'd cards go for pissed off miners that couldn't sustain their mining memory OC because they degraded their chips? Granted I don't know the percentage of this, but it's definitely not 0 from reading through various forums and posts elsewhere.
All I am saying is for people who buy this B-Stock card, or any B-Stock card, it doesn't hurt to test even a mild OC on the Memory.
How many 3090's do you have mining right this moment?
My friend used his 3090 to mine which I helped to put extra heatsink on backplate to keep the memory junction temperature to 90 C degree. However, the card failed within one year showing black screen upon gaming launching or miner starting. I tested with memtestG80 and confirmedVRAM failure. He RMA'ed the card and got a refurbished one.
My friend used his 3090 to mine which I helped to put extra heatsink on backplate to keep the memory junction temperature to 90 C degree. However, the card failed within one year showing black screen upon gaming launching or miner starting. I tested with memtestG80 and confirmedVRAM failure. He RMA'ed the card and got a refurbished one.
Thanks, can you please cite any other place this was previously reported?
24 Comments
Your comment cannot be blank.
Sign up for a Slickdeals account to remove this ad.
Furthermore, for all the alarming talk about vram junction temps, there have been no reported cases of memory failure or degradation. I got a 3080 Ti that came out of the box as an already poor memory OCer, it has not degraded after nearly a year. Heck, HIVE OS didn't even report vram junction temp until 2 months ago.
These cards have 99 problems, but memory temps is not one of them.
It all depends on the card, but my 3080 Ti FTW3 would be in the upper 70's or lower 80's depending on ambient temp. My 3090 would sometimes hit 90c+ due to just having a backplate on the memory connected with pads, and nothing else.
I never said failure, but degradation is a real thing on some 3000 series cards. Simple internet search (pick your search provider) of 3090 VRAM degradation will point you in the correct direction. The FE cards are most at fault hitting over 105c in general regardless of which FE card, but AIB cards can be tricky too. Even across identical cards, temps can fluctuate.
And as an additional note, I replaced the rear pads with a copper plate (for direct contact to the backplate) just for testing sakes, and that dropped my memory junction temps by 6c. This proves the original pads aren't the greatest.
I also added a heatsink on to the backplate afterwards and that dropped it another 10c. After all said and done, was getting in the low 70's for memory junction temp. Did I need to do all of this? Probably not, but I want my card to last a long time. I've attached pics of my mods. 20c+ memory drop when all was said and done.
Where do you think RMA'd cards go for pissed off miners that couldn't sustain their mining memory OC because they degraded their chips? Granted I don't know the percentage of this, but it's definitely not 0 from reading through various forums and posts elsewhere.
All I am saying is for people who buy this B-Stock card, or any B-Stock card, it doesn't hurt to test even a mild OC on the Memory.
Everyone makes mistakes in assuming, so I won't trash talk you because of it.
It all depends on the card, but my 3080 Ti FTW3 would be in the upper 70's or lower 80's depending on ambient temp. My 3090 would sometimes hit 90c+ due to just having a backplate on the memory connected with pads, and nothing else.
I never said failure, but degradation is a real thing on some 3000 series cards. Simple internet search (pick your search provider) of 3090 VRAM degradation will point you in the correct direction. The FE cards are most at fault hitting over 105c in general regardless of which FE card, but AIB cards can be tricky too. Even across identical cards, temps can fluctuate.
And as an additional note, I replaced the rear pads with a copper plate (for direct contact to the backplate) just for testing sakes, and that dropped my memory junction temps by 6c. This proves the original pads aren't the greatest.
I also added a heatsink on to the backplate afterwards and that dropped it another 10c. After all said and done, was getting in the low 70's for memory junction temp. Did I need to do all of this? Probably not, but I want my card to last a long time. I've attached pics of my mods. 20c+ memory drop when all was said and done.
Where do you think RMA'd cards go for pissed off miners that couldn't sustain their mining memory OC because they degraded their chips? Granted I don't know the percentage of this, but it's definitely not 0 from reading through various forums and posts elsewhere.
All I am saying is for people who buy this B-Stock card, or any B-Stock card, it doesn't hurt to test even a mild OC on the Memory.
FarCry 6 got Memory junction temps pretty high for what it's worth. Not sure if any games in the future that are VRAM intensive will.
FarCry 6 got Memory junction temps pretty high for what it's worth. Not sure if any games in the future that are VRAM intensive will.
https://www.google.com/search?q=3...eg
There are none. There's just a bunch if chicken littles misreading datasheets and the tech press (Igor specifically) going off half cocked.
Feel free to mod your card all you want for improved temps, but please stop spreading baseless misinformation.
https://www.google.com/search?q=3...eg
There are none. There's just a bunch if chicken littles misreading datasheets and the tech press (Igor specifically) going off half cocked.
Feel free to mod your card all you want for improved temps, but please stop spreading baseless misinformation.
Six.
You have a mouse? Gotta click the links to read in the forums buddy. Enjoy
- Any GDDR6X failure, period
or
- Any GDDR6X failure due to high temps
or
- Any GDDR6X failure due to mining.
or
- Any measurable GDDR6X degradation
I'll wait.
Sign up for a Slickdeals account to remove this ad.
- Any GDDR6X failure, period
or
- Any GDDR6X failure due to high temps
or
- Any GDDR6X failure due to mining.
or
- Any measurable GDDR6X degradation
I'll wait.
Do you know his preferred mining settings?
What was the make and model of GPU?
Do you know his preferred mining settings?
What was the make and model of GPU?
The card was EVGA 3090 Kingpin. This was only personal experience.