Is my GPU broken.

[Mole_ZA]

Hi guys,

So I have a Palit GTX 1070Ti and it seems to be under performing, I have it at stock clock speeds and it’s running almost the exact same numbers as my friends MSI GTX 1070, I have a newer CPU, DDR4 RAM vs his DDR3 both running 16GB. Is there something I’m missing or do I have an incorrectly binned GPU?

Sent from my iPhone using Tapatalk

 

[VandaL]

Hi guys,

So I have a Palit GTX 1070Ti and it seems to be under performing, I have it at stock clock speeds and it’s running almost the exact same numbers as my friends MSI GTX 1070, I have a newer CPU, DDR4 RAM vs his DDR3 both running 16GB. Is there something I’m missing or do I have an incorrectly binned GPU?

Sent from my iPhone using Tapatalk

Is his Gpu oc'd? Heaven is 99% Gpu ddr4 and a modern gen cpu help with a few fps. Judging from this firestrike bench a oc'd 1070 can marginally beat a 1070ti. Pascals boost is heavily dependant on temps as well. So if his is running cooler it could just be boosting higher.

 

[Porra]

Remember a 1070Ti card is the same card as a normal 1070 that is just slightly higher binned. So if he has a slight oc running in the background or in afterburner and you running stock he'll definitely keep up. Those msi cards also hold their temps really well.

 

[Mole_ZA]

Ok, so I know my card is not running as cool as his is so that could could be an easy justification. I just really thought there would be a more substantial difference between the two, he is also running a mild OC, +100 on core and memory.

Thanks for the insight guys.


Sent from my iPhone using Tapatalk

 

[Mole_ZA]

Also I’m pretty sure I’m using an ex mining card. When playing Ark, the temps go 80+ I’ve increased fan speed but they are noisy, I’ve ordered new fans. Is it worth replacing thermal past as well?


Sent from my iPhone using Tapatalk

 

[PaladinLaw]

Remember a 1070Ti card is the same card as a normal 1070 that is just slightly higher binned. So if he has a slight oc running in the background or in afterburner and you running stock he'll definitely keep up. Those msi cards also hold their temps really well.

Not quite bud. A 1070ti is a cut down 1080, not a beefed up 1070

 

[Porra]

Not quite bud. A 1070ti is a cut down 1080, not a beefed up 1070

Hahahaha dude, that's a bigger debate then team red or team blue. They all within +/- 10% of each other

 

[Mole_ZA]

Not quite bud. A 1070ti is a cut down 1080, not a beefed up 1070

This is why I was expecting a score that would be definitively higher, not huge but noticeable- at least a 10% but I’m putting it down to thermals until it can sort that out.


Sent from my iPhone using Tapatalk

 

[PaladinLaw]

Hahahaha dude, that's a bigger debate then team red or team blue. They all within +/- 10% of each other

True yes, but consider this.

You can't add cudo cores, but you can take them away.

All I'm saying

 

[Mole_ZA]

Hahahaha dude, that's a bigger debate then team red or team blue. They all within +/- 10% of each other

But surely the 10% theory is between cards of the same tier/bin? So the most expensive 1070 vs the cheapest 1070 won’t have a greater than 10% difference. But the 1070Ti isn’t the same card it has more cudo cores and therefore should be hands down better?
That’s just how I understood it.


Sent from my iPhone using Tapatalk

 

[PaladinLaw]

Tell you what, I've got a 1070 aorus, tonight when i get home ill do 2 benches, stock and OC and post them here to compare

 

[PaladinLaw]

Turns put I both had some spare time and was too impatient to wait lol.

So allow me to confuse the matter further.

AMD Ryzen 5 1600x running stock.
1x8gb single channel ddr4 ram running at 2666
1070 aorus extreme.

Temps reached 70 max without OC
73 with a 100mhz OC on core and memory.

Stock test

https://imgur.com/a/PuLlDi6

OC test

https://imgur.com/a/12pnQ3z

Looking at the results, I'd say it's possible the cpu has a role to play here.

If at all possible, plug your gpu into your mates pc and bench again, and compare results

 

[VandaL]

For heaven, the standard bench settings since pascals launch has been 8x AA ,1080p, tesselation extreme.

Run with those settings on both machines to get a good comparative bench result. My 1080tis, SC2 hybrid 161.2fps (8700k 5ghz), auros extreme on a 4770k oc'd to 4.4ghz 154.8fps.
SC2 54c with near silent fan. Auros 60c 100% fans.

 

[PaladinLaw]

For heaven, the standard bench settings since pascals launch has been 8x AA ,1080p, tesselation extreme.

Run with those settings on both machines to get a good comparative bench result. My 1080tis, SC2 hybrid 161.2fps (8700k 5ghz), auros extreme on a 4770k oc'd to 4.4ghz 154.8fps.
SC2 54c with near silent fan. Auros 60c 100% fans.

Will try those settings tonight when I'm home.

Just ran the same settings as OP

 

[etherion]

Mining does not slow a gpu down. If you bought it from someone that mined at work with "free" power it might have run hot so it's silicon life might be shorter. If they paid for power the chances are the card was ran with a power limit and therefore cool so minimum to no effect on life span. From a gpu perspective there is no real difference running math for games vs compute task. Both uses the Cuda drivers provided by nvidia. OC will come down over time due to aging.