Tom's Hardware Superposition Benchmark Thread

[COLGeek]

Something a little different for the thread:

 

[EquineHero]

Something a little different for the thread:

Keeping me busy over here. I'll add my set of cards to the chart soon...you'll see I have a few spare parts... (enough to give 5 PCs 8GB RAM...)

 

[EquineHero]

I think it needs new thermal paste. I half expected it to explode. I could cook an egg on this...let's hope the R9 280X fairs a bit better.

 

[EquineHero]

Wow...this actually performs the same as a mildly overclocked 1050Ti...For the heck of it I ran Passmark as well, not that bad. My 1050Ti got 1458 on Superposition and 7113 on Passmark.

This one actually almost blew up. Apparently a 450W PSU isn't good enough for an overclocked 280X.

Knowing how crippling Superposition is, my GTX 750 is gonna choke on its own blood and soil itself.

 

[jankerson]

I got one that would likely blow up if it actually ran it, way below min specs though, a HD 3870.

Trashed my old HD 5850 awhile back so I don't have that one anymore.

 

[EquineHero]

I got one that would likely blow up if it actually ran it, way below min specs though, a HD 3870.

Trashed my old HD 5850 awhile back so I don't have that one anymore.

Oh fish that's old. That's like 8800GT era old.

I ran my GTX 580, 1.5GB vRAM is severely killing the score and I want to make an exception for 1080p Medium in this case. Superposition was basically a nervous slideshow. I'm posting Passmark scores on these older vRAM limited cards to give you a better idea of how they perform when vRAM isn't an issue. I'm waiting on MSI Afterburner to finish and then I'm gonna run the benchmark again before updating the table. I think if you need to run the benchmark at 1080p Medium, chop the score by 35% rounded up and we can make that the 1080p Extreme score if results are consistent enough between vRAM limited cards.

If I had the GTX 580 3GB, vRAM wouldn't be an issue. If someone does have one around, please benchmark it!

 

[jankerson]

I got one that would likely blow up if it actually ran it, way below min specs though, a HD 3870.

Trashed my old HD 5850 awhile back so I don't have that one anymore.

Oh fish that's old. That's like 8800GT era old.

I ran my GTX 580, 1.5GB vRAM is severely killing the score and I want to make an exception for 1080p Medium in this case. Superposition was basically a nervous slideshow. I'm posting Passmark scores on these older vRAM limited cards to give you a better idea of how they perform when vRAM isn't an issue. I'm waiting on MSI Afterburner to finish and then I'm gonna run the benchmark again before updating the table. I think if you need to run the benchmark at 1080p Medium, chop the score in half rounded up and we can make that the 1080p Extreme score if results are consistent enough between vRAM limited cards.

If I had the GTX 580 3GB, vRAM wouldn't be an issue. If someone does have one around, please benchmark it!

Surprised we haven't seen more people run this.

But then 1080P Extreme is the hardest setting.

The 4K preset is nothing compared to that.

 

[EquineHero]

I got one that would likely blow up if it actually ran it, way below min specs though, a HD 3870.

Trashed my old HD 5850 awhile back so I don't have that one anymore.

Oh fish that's old. That's like 8800GT era old.

I ran my GTX 580, 1.5GB vRAM is severely killing the score and I want to make an exception for 1080p Medium in this case. Superposition was basically a nervous slideshow. I'm posting Passmark scores on these older vRAM limited cards to give you a better idea of how they perform when vRAM isn't an issue. I'm waiting on MSI Afterburner to finish and then I'm gonna run the benchmark again before updating the table. I think if you need to run the benchmark at 1080p Medium, chop the score in half rounded up and we can make that the 1080p Extreme score if results are consistent enough between vRAM limited cards.

If I had the GTX 580 3GB, vRAM wouldn't be an issue. If someone does have one around, please benchmark it!

Surprised we haven't seen more people run this.

But then 1080P Extreme is the hardest setting.

The 4K preset is nothing compared to that.

Well the results for the GTX 580 are in, and they aren't very pretty so I ran 1080p Medium.

4033*0.35= 1412 (35%), I think that's fair considering how the 270x/280x performed.

Yo moderators, you think that's fair? Even the Passmark scores are consistent for a card of its power.

 

[EquineHero]

Surprised we haven't seen more people run this.

But then 1080P Extreme is the hardest setting.

The 4K preset is nothing compared to that.

The amount of artifacting I experienced made me extremely uncomfortable. But those temps tho! Unfortunately adding more RAM did not help my score, neither did overclocking my Xeon or upgrading to the Z77 chipset...Only the water cooler I installed...

 

[Ilkan Akten]

This is the max overclock I can get stable

GPU:
Name:GeForce GTX 1080 (GP104)
Vendor:EVGA
Memory:8192 MB
Memory clock (max):6075 MHz
Core clock (max):2138 MHz
Temp min/max:32 / 61 C°
Driver:382.05

https://benchmark.unigine.com/results/rid_86a8bca22f074e0182317e0f52d29a0c

 

[EquineHero]

This is the max overclock I can get stable

GPU:
Name:GeForce GTX 1080 (GP104)
Vendor:EVGA
Memory:8192 MB
Memory clock (max):6075 MHz
Core clock (max):2138 MHz
Temp min/max:32 / 61 C°
Driver:382.05

https://benchmark.unigine.com/results/rid_86a8bca22f074e0182317e0f52d29a0c

Please read the rules on the first post. Your "score" hasn't been added due to your failure to follow the rules.

 

[bjornl]

This benchmark seems to be bandwidth sensitive more than anything else. I was running on a 5830k at 4.5. Dropped it to 4389 (104.5x42) and it improved slightly. I overclocked the video card 62mhz core and 550mhz ram and it jumped a bit more.

My system ram is currently 'crippled'; one side of the ram sockets are bad due to shoddy ASUS QA, so running single channel. I wonder if this will improve when I swap in the replacement MB and enable quad channel again......

 

[EquineHero]

This benchmark seems to be bandwidth sensitive more than anything else. I was running on a 5830k at 4.5. Dropped it to 4389 (104.5x42) and it improved slightly. I overclocked the video card 62mhz core and 550mhz ram and it jumped a bit more.

My system ram is currently 'crippled'; one side of the ram sockets are bad due to shoddy ASUS QA, so running single channel. I wonder if this will improve when I swap in the replacement MB and enable quad channel again......

Are you bozi12 on the leaderboard? they also have a 5820K.

I don't think it's as sensitive as you believe, my 1070 is running in PCI 2.0 mode and I was still able to get fairly high on the leaderboard, I think my 1070 is lower binned because I haven't been able to get much higher than 3900 without severe artifacting.

 

[bjornl]

This benchmark seems to be bandwidth sensitive more than anything else. I was running on a 5830k at 4.5. Dropped it to 4389 (104.5x42) and it improved slightly. I overclocked the video card 62mhz core and 550mhz ram and it jumped a bit more.

My system ram is currently 'crippled'; one side of the ram sockets are bad due to shoddy ASUS QA, so running single channel. I wonder if this will improve when I swap in the replacement MB and enable quad channel again......

Are you bozi12 on the leaderboard? they also have a 5820K.

I don't think it's as sensitive as you believe, my 1070 is running in PCI 2.0 mode and I was still able to get fairly high on the leaderboard, I think my 1070 is lower binned because I haven't been able to get much higher than 3900 without severe artifacting.

I am not bozi12 an I don't have a 5820k, I have a 5930k. My 980ti would only OC to 1480 (a 60mhz increase). My CPU was only a average overclocker. Before my fiddling mentioned above I also got around 3900. After I got back to back runs of around 4220.

 

[EquineHero]

I am not bozi12 an I don't have a 5820k, I have a 5930k. My 980ti would only OC to 1480 (a 60mhz increase). My CPU was only a average overclocker. Before my fiddling mentioned above I also got around 3900. After I got back to back runs of around 4220.

Feels good to be on top, baby!

SLI GTX 1070s kick major butt in Superposition, double memory bandwidth, great scaling (greater than 200%)

 

[Rogue Leader]

Hah, its about time someone beat me. Too bad it took two GPUs to do it

 

[EquineHero]

Hah, its about time someone beat me. Too bad it took two GPUs to do it

yeah I paid a little over $800 for both, and I feel kinda stupid because one 1070 was $380 on Amazon and the other was $460 at Best Buy and I lost my dang receipt.

I mean the 1070 is definitely not a bad GPU, while I wouldn't say that the score scaled because of the scoring algos, the framerate was practically double vs single GPU.

My old Corsair CX600 is powering these things like a champ too, largely because my CPU draws less power than an equivalent i7.

Now it's your turn!

 

[Rogue Leader]

LOL I don't even know where I would get my hands on another liquid cooled Vega 64, they are sold out everywhere (and when available they are $800). I'd also need a significantly larger PSU, as is I am just about maxing out my 650. Nah, I'm good with second place !

 

[EquineHero]

LOL I don't even know where I would get my hands on another liquid cooled Vega 64, they are sold out everywhere (and when available they are $800). I'd also need a significantly larger PSU, as is I am just about maxing out my 650. Nah, I'm good with second place !

I added a light overclock to core and memory. HUGE improvement - my FPS increased from a single GPU at a factor of OVER 200%.

Sell your Vega and get two 1070s, I mean seriously, look at this!

Edit: I should mention that this score is better than almost all of the pre-Titan V scores - coming at 14th place WORLDWIDE, and minus the Titan V the score is 3rd worldwide. Good luck beating me with a 1080Ti hahahaha

 

[Rogue Leader]

That is an insane score. Tempting but my system runs well as is and I like the pretty colors lol

 

[EquineHero]

That is an insane score. Tempting but my system runs well as is and I like the pretty colors lol

But I have pretty colors too. The higher end GTX 1070s have RGB lights.

 

[COLGeek]

Two looks at RX480s in action.

Win 10 Pro x64:

Windows Server Essentials 2016:

 

[EquineHero]

Two looks at RX480s in action.

Win 10 Pro x64:

Windows Server Essentials 2016:

Put them...put them in Crossfire...

Those are both below Jankerson's benchmark...

If you're having clock problems I feel bad for you son, I got 99 problems and a bench ain't one

Also, because of the rules I've put in place, only one of those will be added, mainly Rule #5 and Rule #9. I know they're both different aftermarket variants, however the rules still stand. If I had an EVGA GTX 1080 and an MSI GTX 1080 i'd only add the better score.

 

[COLGeek]

CF isn't needed. I was just posting for others, not the chart.

The one in my server (the second one) is primarily there to crunch numbers (BOINC/SETI). The other rig is my daily driver PC.

No OCs on anything (including my game rig). Might do some tweaking over the holidays to see what I can squeeze out them.

 

[logainofhades]

That is an insane score. Tempting but my system runs well as is and I like the pretty colors lol

But I have pretty colors too. The higher end GTX 1070s have RGB lights.

I wish my cards had that, as the Neutron shows off the graphics card.