Question Purchased "refurbished" Palit rtx 4090 few days ago but performing poorly...

[badtimes]

Hey guys - new user here with a question.

Few days ago I got my Palit RTX 4090 delivered and put it into my PC myself, even though I might not be the most experienced to do so. While doing that, I also had to upgrade my 600w PSU from BeQuiet, and "throw out" my gainward gtx 1080.

I've tried running some benchmarks - most importantly Userbenchmark.com ( https://www.userbenchmark.com/UserRun/61318246 ) and I've noticed my GPU is performing very, very poorly.

Now my question is, does anyone of you with a little more insight than me can point me to a benchmark where I can compare solely the GPU to other users GPUs - preferably the same? I must admit I am kind of scared that I've done something wrong - or, since my GPU is "b-ware" (sent back from someone else and then sold to me for $100 less), it's not working like it should.

What also worries me is, that I've seen online that some rtx 4090s come with a 4-connector 12vhpwr cable adapter, while mine only has 3 out of the box so I'm wondering if I should get myself a 4-connector cable adapter and throw that on instead, so my 4090 can grab more power? I've also installed Palits ThunderMaster and out of the box it doesn't allow me to set power % to 120%, which I could do with my 1080 before so I'm worried it's not getting enough power and that is why it's performing poorly.

So if there's anybody that can help me gain some insight, please let me know!


Thank you

 

[kerberos_20]

Hey guys - new user here with a question.

Few days ago I got my Palit RTX 4090 delivered and put it into my PC myself, even though I might not be the most experienced to do so. While doing that, I also had to upgrade my 600w PSU from BeQuiet, and "throw out" my gainward gtx 1080.

I've tried running some benchmarks - most importantly Userbenchmark.com ( https://www.userbenchmark.com/UserRun/61318246 ) and I've noticed my GPU is performing very, very poorly.

Now my question is, does anyone of you with a little more insight than me can point me to a benchmark where I can compare solely the GPU to other users GPUs - preferably the same? I must admit I am kind of scared that I've done something wrong - or, since my GPU is "b-ware" (sent back from someone else and then sold to me for $100 less), it's not working like it should.

What also worries me is, that I've seen online that some rtx 4090s come with a 4-connector 12vhpwr cable adapter, while mine only has 3 out of the box so I'm wondering if I should get myself a 4-connector cable adapter and throw that on instead, so my 4090 can grab more power? I've also installed Palits ThunderMaster and out of the box it doesn't allow me to set power % to 120%, which I could do with my 1080 before so I'm worried it's not getting enough power and that is why it's performing poorly.

So if there's anybody that can help me gain some insight, please let me know!


Thank you

3x8pin should have 450watts available to your GPU
whats your gpu power consumption under load?
you can check it with gpu-z

board powerdraw should be close to 450watts

4x8pin would allow you to overclock your gpu up to 600 watts

 

[badtimes]

Okay this is while stress testing with FurMark

Does that look okay? Would be nice if I had a way to compare to other ppls 4090s while doing this

 

[kerberos_20]

that means youre CPU bottlenecked on userbenchmark gpu test
run unigine superposition or 3dmark

 

[badtimes]

that means youre CPU bottlenecked on userbenchmark gpu test
run unigine superposition or 3dmark

Yeah okay I edited the other post by accident instead of posting a new reply.
do you know any benchmarking software that allows me to compare to other 4090s similar to userbenchmark?

I'll try 3dmark next

 

[kerberos_20]

Yeah okay I edited the other post by accident instead of posting a new reply.
do you know any benchmarking software that allows me to compare to other 4090s similar to userbenchmark?

I'll try 3dmark next

3d mark will let you compare gpu, but i can tell even without comparison that you wont hit 36k gpu score which acording to 3dmark rtx 4090 offers

you gpu is power limited for sure, at 2.5GHz hitting 450watts, you should be able to reduce voltage and increase clock to something around 2.7GHz (msi afterburner curve overclocking)
now for 3dmark 36k score, you would need 4x8pin to reach 3GHz

 

[badtimes]

3d mark will let you compare gpu, but i can tell even without comparison that you wont hit 36k gpu score which acording to 3dmark rtx 4090 offers

you gpu is power limited for sure, at 2.5GHz hitting 450watts, you should be able to reduce voltage and increase clock to something around 2.7GHz (msi afterburner curve overclocking)
now for 3dmark 36k score, you would need 4x8pin to reach 3GHz

This is my score on 3dmark - below average unfortunately. I don't want to break my 4090 or burn down my entire PC by overclocking to a level the connector can't withstand - I don't even know if nVidias connector has been "fixed".

I guess I'll just stick to this being below average, which is very sad.

 

[badtimes]

At least it doesn't look like I connected it incorrectly, right?

 

[kerberos_20]

This is my score on 3dmark - below average unfortunately. I don't want to break my 4090 or burn down my entire PC by overclocking to a level the connector can't withstand - I don't even know if nVidias connector has been "fixed".

I guess I'll just stick to this being below average, which is very sad.

its not really below average, if you look on that chart, youre pretty much where most people are
that green line is your score

 

[1LiquidPC]

Your CPU by today's standards will be limiting performance as well. I had a 9900K myself awhile back, but when paired with a 4090, it's definitely limiting the GPU performance across 1080P, 1440P, and 4K to a degree. As you had mentioned, the adapter being 3X to 12-pin vs 4X, it's limiting the power available for raising the limit to 120%

 

[Bob.B]

Hey guys - new user here with a question.

Few days ago I got my Palit RTX 4090 delivered and put it into my PC myself, even though I might not be the most experienced to do so. While doing that, I also had to upgrade my 600w PSU from BeQuiet, and "throw out" my gainward gtx 1080.

I've tried running some benchmarks - most importantly Userbenchmark.com ( https://www.userbenchmark.com/UserRun/61318246 ) and I've noticed my GPU is performing very, very poorly.

Now my question is, does anyone of you with a little more insight than me can point me to a benchmark where I can compare solely the GPU to other users GPUs - preferably the same? I must admit I am kind of scared that I've done something wrong - or, since my GPU is "b-ware" (sent back from someone else and then sold to me for $100 less), it's not working like it should.

What also worries me is, that I've seen online that some rtx 4090s come with a 4-connector 12vhpwr cable adapter, while mine only has 3 out of the box so I'm wondering if I should get myself a 4-connector cable adapter and throw that on instead, so my 4090 can grab more power? I've also installed Palits ThunderMaster and out of the box it doesn't allow me to set power % to 120%, which I could do with my 1080 before so I'm worried it's not getting enough power and that is why it's performing poorly.

So if there's anybody that can help me gain some insight, please let me know!


Thank you

Start with.
Reboot.
Wait a few mins.
Run ubm with the browser closed.
Post a link to the results page.

 

[badtimes]

Your CPU by today's standards will be limiting performance as well. I had a 9900K myself awhile back, but when paired with a 4090, it's definitely limiting the GPU performance across 1080P, 1440P, and 4K to a degree. As you had mentioned, the adapter being 3X to 12-pin vs 4X, it's limiting the power available for raising the limit to 120%

Ha I didn't think my i9 9900k would bottleneck my GPU as it's been a top notch CPU not so long ago. Makes sense though.

Now I've downloaded MSI Afterburner and had the OC-Checker run and OC my 4090 automatically so I gained a bit more performance in the 3DMark benchmark I did before. I'm a little above average so I guess I'm happy with that.

Especially now that I know I didn't <Mod Edit> up connecting my 4090 or knowing that it isn't "broken".

 

[Avro Arrow]

Well, for one thing, I wouldn't be using LoserBenchmark. If you want to properly benchmark a GPU, you want to use something like 3DMark or Unigine Superposition.