Question 6800 vs 6800XT vs N32

[biostud]

I'm planning to upgrade my system feb/march when the 7800X3D is released so I can decide which 8-core CPU makes most sense to me. I have an Odyssey G9, but I mostly play older games, and world of tanks and no interest in raytracing (at current pricing and performance). The 6800 series seems to offer bester price/performance atm, an N32 arrival is unknown. So what would you do? Buy now before stock is gone, wait and maybe miss out, or something else? I really dont want to spend much more than $800 in Danish prices for a video card.


(Prices for comparison)
3080 10GB ($999)
7900XT ($1142)
4080 ($1570)

 

[igor_kavinski]

Yeah, it's likely something is wrong with the driver setup or 3dmark.

I would think the latter. 3DMark won't run for me even though I PAID $5 for it. Buggy software.

 

[PJVol]

Buggy software

Once there was a bug when it failed to verify the results, and it fixed itself after reinstalling the driver.

 

[biostud]

I couldn't Uninstaller the nvidia drivers, so I had to use DDU.

So if it doesn't work properly that could be the culprit, but I'm not going to spend a lot of time getting 3dmark to run, as it not a game I plan to play.

 

[biostud]

So did a little testing on BL3
@200W avg. FPS 66.52
@280W avg. FPS 71.25

So 7% faster with 40% more power consumption, or 6.6% slower with 28.5% less power used.

Or 3W/fps vs 4W/fps

@200W avg. FPS 66.52 (actual GPU voltage ~950mV, max frequency 2300Mhz, Voltage 1050 mV)
@240W + 20% avg FPS 69.51 +4,5% (actual GPU voltage ~1015mV, max frequency 2400Mhz, Voltage 1050 mV)
@280W + 40% avg. FPS 71.25 + 7,1% (actual GPU voltage ~1150mV, max frequency 2500Mhz, Voltage 1050 mV)
@310W + 55% avg. FPS 72.54 + 9,0% (actual GPU voltage ~1150mV, max frequency 2600Mhz, Voltage 1150 mV)

I'll stick with the 2300Mhz setting for now

 

[Chicken76]

So did a little testing on BL3
@200W avg. FPS 66.52
@280W avg. FPS 71.25

So 7% faster with 40% more power consumption, or 6.6% slower with 28.5% less power used.

Or 3W/fps vs 4W/fps

What do you expect? When the competition melts cables in fear of being 2 fps behind in benchmarks you gotta make some compromises.

 

[biostud]

What do you expect? When the competition melts cables in fear of being 2 fps behind in benchmarks you gotta make some compromises.

I didn't really expect that it was possible to save that much power, without sacrificing too much performance. Overall (and I kinda knew this) I just think it is sad how much the lock down their bios for modding and voltage regulation. I can understand why EVGA left the market, as all the tinkering has effectively been killed, and that is ultimately what made this hobby really fun.

 

[biostud]

Will instability of a videocard always result in the drivers crashing or can it also just be the game crashing and everything else just working as usual?

 

[Hitman928]

Will instability of a videocard always result in the drivers crashing or can it also just be the game crashing and everything else just working as usual?

Unstable video card could cause a game crash without the drivers crashing. It can even cause the OS to crash, though nowadays, operating systems are usually more graceful about it and will at least point to the graphics card (or graphics card driver crashing) as the reason for the crash before rebooting.

 

[igor_kavinski]

Sometimes the driver will crash and get restarted without rebooting the whole OS. Saw this happen with Nvidia drivers.

 

[biostud]

I played Darktide yesterday and had a couple of crashes, but I also experienced that before. And it throwing me server errors, so whether it's video card stability or just buggy mess....

 

[PJVol]

I played Darktide yesterday and had a couple of crashes, but I also experienced that before. And it throwing me server errors, so whether it's video card stability or just buggy mess....

Most of the time driver resets overdrive settings, i.e. mem and gfx clocks, voltage, etc.

 

[In2Photos]

I got some time to play with my 6800XT today and was surprised by some results. I still plan to tune some more but wanted to share my findings and see if anyone else experienced the same. I used Unigine Heaven 4.0 benchmark at 1080p Ultra, with AA-x8, not full screen, all other settings left at default.

Using the default settings in Adrenaline I got the following scores and readings

334 fps, 8412 score @ 2400 MHz GPU clock, 1986 MHz mem clock, 1.103V, 253W, 1380RPM, temps of 65/84C

I then benched it using all of the 1-click settings in Adrenaline and for the most part the results are all within a few points of each other. I decided to see how far I could take the card while trying to keep thermals near stock and ended up with these results.

358 fps, 9010 score @ 2624 MHz GPU clock, 2138 MHz mem clock, 1.102V, 269W, 2000RPM, temps of 66/87

I'm shocked that the power didn't go through the roof with over 200 MHz overclock. Screenshot of the adrenaline settings.

 

[PJVol]

if anyone else experienced the same

Why don't you just post a screenshot of the result since it's not clear what is meant by "all other settings at default".

 

[In2Photos]

Why don't you just post a screenshot of the result since it's not clear what is meant by "all other settings at default".

Here you go.

 

[PJVol]

I'm shocked that the power didn't go through the roof with over 200 MHz overclock.

Why should it when you limit it at 274W, thus determining the operation frequency range, and not vice versa.
Also, some observations:

* Better to use "Extreme" preset (1080p, Quality - Ultra, Tesselation - Extreme) to minimize CPU bottleneck.
At least there's a thread at TPU with a bunch of results with these settings at 1080p and 1440p.

* Setting Min.frequency gives you nothing.

* Old DX11 benchmark is really bad choice to experiment with the perf/power on Radeons (Vulkan or DX12 is still preferable)

* my scores:

240 fps - 255W stock
249 fps - 255W tuned
254 fps - 275W tuned

So, yes, tuning the card alone gives way more perf. within the same power limit, than just power budget increase.

... ... ...

..Stock preset .......... OC @ 255W.......... OC @ PL +8%
..... .

Tesselation "disabled" (stock)
...
..

 

[Joe NYC]

View attachment 74806
Now I just need to get the kids tucked in...

Looks like you already pulled the trigger.

I will have to decide what to get or find a way to wait for N32 (by moving a 6600 XT from my work computer, temporarily)...

But I am intrigued by the new price of $699 for 6950 XT directly from AMD.

The performance of full Navi 32 with 16 GB of memory should be very close to the 6950 XT, and price will also be quite similar. Here are a couple of graphs by AMD to compare their cards:



And price per FPS chart:

 

[biostud]

I got some time to play with my 6800XT today and was surprised by some results. I still plan to tune some more but wanted to share my findings and see if anyone else experienced the same. I used Unigine Heaven 4.0 benchmark at 1080p Ultra, with AA-x8, not full screen, all other settings left at default.

Using the default settings in Adrenaline I got the following scores and readings

334 fps, 8412 score @ 2400 MHz GPU clock, 1986 MHz mem clock, 1.103V, 253W, 1380RPM, temps of 65/84C

I then benched it using all of the 1-click settings in Adrenaline and for the most part the results are all within a few points of each other. I decided to see how far I could take the card while trying to keep thermals near stock and ended up with these results.

358 fps, 9010 score @ 2624 MHz GPU clock, 2138 MHz mem clock, 1.102V, 269W, 2000RPM, temps of 66/87

I'm shocked that the power didn't go through the roof with over 200 MHz overclock. Screenshot of the adrenaline settings.

View attachment 75392

With my very short experience with tinkering around, you will experience different clock speeds and power consumption depending on which game/benchmark you run. Heaven doesn't seem to be very demanding of the GPU. If you run a modern game with full screen, I'm sure your clock speeds will not be as high.

But overall, it seems like our findings of performance vs power consumption is very similar.

 

[biostud]

Looks like you already pulled the trigger.

I will have to decide what to get or find a way to wait for N32 (by moving a 6600 XT from my work computer, temporarily)...

But I am intrigued by the new price of $699 for 6950 XT directly from AMD.

The performance of full Navi 32 with 16 GB of memory should be very close to the 6950 XT, and price will also be quite similar. Here are a couple of graphs by AMD to compare their cards:

View attachment 75417

And price per FPS chart:

View attachment 75418

I think you will have to wait a long time for the N32, and to me this chart confirms this suspicion. So I'm a happy camper, now I'll just need the 7800X3D and Fractal North to be available, to do a total upgrade.

 

[Gideon]

I think you will have to wait a long time for the N32, and to me this chart confirms this suspicion. So I'm a happy camper, now I'll just need the 7800X3D and Fractal North to be available, to do a total upgrade.

Having played around with my father's new 7700X and owning a 5800X3D for a month, I have to say, the 7800X3D really looks like the best all-around CPU to own. I would have waited for one myself, but as I was already on AM4 platform, going for a discounted 5800X3D made a lot more sense.

If the 7800X3D behaves even slightly similarly to the 5800X3D though, be sure to undervolt it with PBO2. You will be getting crazy-low gaming power consumption (e.g. ~45-55 W) and 100-200Mhz higher all-core boost clocks than without it. Without undervolt the 5800X3D hit it's 90° thermal wall very quickly, even with good cooling (while 5800X did not). I can't imagine 7800X3D being much different considering the 7700X is also glued around 90° when stock.

BTW I see, you are still rocking the i7 in your signature. That's impressive value out of the rig, nearly 10 years!

 

[biostud]

Having played around with my father's new 7700X and owning a 5800X3D for a month, I have to say, the 7800X3D really looks like the best all-around CPU to own. I would have waited for one myself, but as I was already on AM4 platform, going for a discounted 5800X3D made a lot more sense.

If the 7800X3D behaves even slightly similarly to the 5800X3D though, be sure to undervolt it with PBO2. You will be getting crazy-low gaming power consumption (e.g. ~45-55 W) and 100-200Mhz higher all-core boost clocks than without it. Without undervolt the 5800X3D hit it's 90° thermal wall very quickly, even with good cooling (while 5800X did not). I can't imagine 7800X3D being much different considering the 7700X is also glued around 90° when stock.

BTW I see, you are still rocking the i7 in your signature. That's impressive value out of the rig, nearly 10 years!

Yeah I'm really looking forward to upgrading. Intel kind of made it easy not wanting to upgrade for a long time, and when AMD finally returned to the scene, I had three small girls that required attention day and night. So finally the time has come, for a total upgrade from a rig bought sep' 2014.

 

[In2Photos]

Why should it when you limit it at 274W, thus determining the operation frequency range, and not vice versa.
Also, some observations:

* Better to use "Extreme" preset (1080p, Quality - Ultra, Tesselation - Extreme) to minimize CPU bottleneck.
At least there's a thread at TPU with a bunch of results with these settings at 1080p and 1440p.

* Setting Min.frequency gives you nothing.

* Old DX11 benchmark is really bad choice to experiment with the perf/power on Radeons (Vulkan or DX12 is still preferable)

* my scores:

240 fps - 255W stock
249 fps - 255W tuned
254 fps - 275W tuned

So, yes, tuning the card alone gives way more perf. within the same power limit, than just power budget increase.

View attachment 75403... View attachment 75416... View attachment 75400... View attachment 75404

..Stock preset .......... OC @ 255W.......... OC @ PL +8%
View attachment 75405 ...View attachment 75406.. .View attachment 75415

Tesselation "disabled" (stock)
...View attachment 75407
..

Sorry it took so long to get back to this. I didn't realize the slider had been reset. I thought it was still maxed out. I was able to run it again yesterday and the system still only peaked at 279W, but achieved 2670MHz this time. What I meant from my prior post about this behavior was that I assumed that this GPU would behave similarly to a CPU or Nvidia GPU in that when you push it above stock you get very little performance increase for a huge jump in power. Seeing 250MHz in additional clocks for ~20W is pretty good IMO. And I was still able to achieve that with decent thermals and moderate fan speeds. I have no desire to keep it there though, just wanted to see how far it will go.

With my very short experience with tinkering around, you will experience different clock speeds and power consumption depending on which game/benchmark you run. Heaven doesn't seem to be very demanding of the GPU. If you run a modern game with full screen, I'm sure your clock speeds will not be as high.

But overall, it seems like our findings of performance vs power consumption is very similar.

I've played a couple of games and ran the Division 2 benchmark using a more "optimized" settings. I like using Heaven for it's repeatability and the tangible score, plus it takes about 4.5 minutes which is longer than game benchmarks so things normalize a little better. I was very happy with the gameplay though. A much smoother experience than my previous system and with much higher graphics settings to boot!

I settled on the following for Adrenaline for now.
Max freq - 2400MHz
Voltage - 1020
Mem clock - +150
Mem timings - Fast
Power limit - 0
Fan tuning - Max speed 55% (the PC is so quiet unlike my previous rig)

In Heaven this gave me 96% (321/334 fps) of the performance but at 70% (176/253W) of the power! Temps were lower and fan speeds were slower as well.

 

[biostud]

Sorry it took so long to get back to this. I didn't realize the slider had been reset. I thought it was still maxed out. I was able to run it again yesterday and the system still only peaked at 279W, but achieved 2670MHz this time. What I meant from my prior post about this behavior was that I assumed that this GPU would behave similarly to a CPU or Nvidia GPU in that when you push it above stock you get very little performance increase for a huge jump in power. Seeing 250MHz in additional clocks for ~20W is pretty good IMO. And I was still able to achieve that with decent thermals and moderate fan speeds. I have no desire to keep it there though, just wanted to see how far it will go.

I've played a couple of games and ran the Division 2 benchmark using a more "optimized" settings. I like using Heaven for it's repeatability and the tangible score, plus it takes about 4.5 minutes which is longer than game benchmarks so things normalize a little better. I was very happy with the gameplay though. A much smoother experience than my previous system and with much higher graphics settings to boot!

I settled on the following for Adrenaline for now.
Max freq - 2400MHz
Voltage - 1020
Mem clock - +150
Mem timings - Fast
Power limit - 0
Fan tuning - Max speed 55% (the PC is so quiet unlike my previous rig)

In Heaven this gave me 96% (321/334 fps) of the performance but at 70% (176/253W) of the power! Temps were lower and fan speeds were slower as well.

If you run with these settings in a game, and use Radeon overlay to monitor your power draw, what values do you see?

For me in the BL3 Benchmark it was ~240W: 2400Mhz, 1050mV, Memclock+100, timings fast, so pretty far from the 176W you are getting in Heaven. I'm just wondering if you're seeing the same power draw when you are actually gaming.

 

[In2Photos]

If you run with these settings in a game, and use Radeon overlay to monitor your power draw, what values do you see?

For me in the BL3 Benchmark it was ~240W: 2400Mhz, 1050mV, Memclock+100, timings fast, so pretty far from the 176W you are getting in Heaven. I'm just wondering if you're seeing the same power draw when you are actually gaming.

Trying to get a screenshot in game with the overlay seems impossible for some reason. So instead I captured everything immediately after the benchmark in Division 2. I ran this at 1080p Ultra, full screen and only turned v-sync off for max frame rate.

Default


My "176W" Optimized profile


While watching the benchmark I noticed a peak of 195W in the overlay, but most of the benchmark it stayed below 180W. HWinfo shows average clock of 2255 which is a little low due to right before and just after the benchmark being included. I noticed it hovering around 2330MHz during the benchmark. Average power is 171W with a peak of 192W.

 

[biostud]

Trying to get a screenshot in game with the overlay seems impossible for some reason. So instead I captured everything immediately after the benchmark in Division 2. I ran this at 1080p Ultra, full screen and only turned v-sync off for max frame rate.

Default
View attachment 75584

My "176W" Optimized profile
View attachment 75585

While watching the benchmark I noticed a peak of 195W in the overlay, but most of the benchmark it stayed below 180W. HWinfo shows average clock of 2255 which is a little low due to right before and just after the benchmark being included. I noticed it hovering around 2330MHz during the benchmark. Average power is 171W with a peak of 192W.

As far as I can see, the only difference is that you can do 1020mV, which gives you the same vcore I can get @ 2300Mhz. If I try that it is not stable, and the power consumption is still not close to yours. Maybe it is because I run on an old system?


1020mV (Unstable):

 

[biostud]

Maybe it is related to my monitor, when I'm in windows the power consumption is 30-35W @ 240Hz?

So droping from 240Hz to 120Hz on my display and my idle power, halved. I could also see that vram speeds dropped from 908Mhz to ~192Mhz....