Uninstalling GeForce 441.87 drivers, downloading Adrenaline 20.5.1...
- Thread starter Lew Zealand
- Start date
Pffft, no - 9200 is bigger than 5700. It’s going to be 4500 slower.
I suppose I'll write a followup after 2 weeks with it. First, what did I do? I sidegraded a GTX 1080 to a 5600XT. They're pretty much the same speed so the experience shouldn't be too different in-game except for any problems encountered from the card platform or the card itself.
Why did I do it? 'Cuz. Yeah, pretty much that's it. I had a lot of extra BezosBucks I hadn't originally planned to have which I was gonna spend on a ~$130 RX 570 and actually had enough for a decent current gen AMD card, the 5600XT.
The actual cards? I replaced a PNY GTX 1080 OC with a Sapphire Pulse 5600 XT.
The PNY is a bottom of the barrel 1080 with a good GPU and memory but a crud cooler design. The 1080 is a nominal 180W card and that can be extended 20% to 216W but the cooler is barely a 150W capable unit and frankly I'd call it a 120W unit. One of the 2 fans started rattling btw 65% and 80% fan speed after a few months so I set my manual fan curve in MSI Afterburner to hop directly from 65% to 80% at 74C. Lite & med gaming keeps below 74C as I manually undervolted the card (usually to 1822MHz and 0.85v) and if I went to 1911MHz and 0.90v (in Control or Jedi:FO), I'd usually set the fans to 80 or 85% manually.
All that and more to compensate for a crap cooler, though I got the card for $440 after RTX started shipping so it was a decent deal. And memory OCs to max 12GBps so there's great memory in this unit. I always intended it to be run undervolted and not madly overclocked as it was replacing a 1060 and even at 1822MHz is way faster. Frankly, if it hadn't been for the fan thing I'd be happy with it.
However I wanted to see how it is in Radeonville so I got the Sapphire Pulse 5600XT, which has been favorably reviewed everywhere so I knew I was getting probably the best deal on any midrange card available right now. Which is good as this is really just a fun buy.
Well, the Pulse's build is everything the PNY isn't, quite hefty with a taller cooler (still fits in my cheap case), and since the 5600XT is a purposefully limited unit so it doesn't compete with the 5700 too much (though it does and eliminates the 5700 as a decent buy), I have trouble getting it to run warmer than 62C. In fact on most games, it's rummaging around 56C and Control barely gets it to that 63C (in a ~26C room). I undervolt it to 0.918v at 1775MHz so it may run warmer at it's stock voltage which is somewhere around 0.98v.
It performs about the same in every game as the 1080 with no noticeable differences, though it does benchmark lower in Dirt Rally (about 120fps to the 1080's 135 fps or so @1440p) and it may be a frame or 2 faster in Control (both in the low 50s @1440p), but neither feel any different while playing. The biggest difference is in vanilla Minecraft of all things. The Pulse 5600XT gets about half the FPS of the 1080 though both are heavily bottlenecked by Minecraft's 3D implementation as both usually run at less than 20% utilization (with a i5-8400 which itself is only running at about 30% util). MC is crap like that. Both run better with OptiFine installed, at around 70FPS @1440p with long draw distances and mods.
So, the black screens or similar. I can make the 5600XT do a black screen easily and also avoid them entirely. Here's how: Afterburner. If I enable AB to have hardware-level access to the 5600XT, it outputs black screens on occasion during high utilization times (benchmarks, etc), though more rarely in-game. Don't give AB that hardware-level access and no black screens. Maybe there's another way to induce them using other software but I don't run anything else that does this. I can still use AB sans HW-level access for it's FPS counter with no black screens.
If you're coming to the Radeon side from the GTX/RTX side, and you're an enthusiast, you probably depend on AfterBurner to squeeze your GTX/RTX to it's limits and you simply need to let that go and use AMD's Radeon software to do the same. I have about 3 years squeezing blood from my 4 GTX cards with differing capabilities using AfterBurner so after 2 weeks in, I'm still in the learning phase and here's what I see:
Some of the numerical settings in the Radeon software do not correlate 1:1 with what TPU's GPU-Z reports or even the documentation for the card describe. WTF?? These are simply numbers and should be dead simple to get exactly correct. You set the card to 1820 MHz (max for the 5600XT) in Radeon SW and it tops out around 1775 MHz. All other speeds scale accordingly, it's not just a 45MHz buzzcut off the top. You set the fans to 35% and they run at 50% and also scale like this up and down. WTF? This is not rocket surgery. OK, the memory speed and voltages are correctly set and manageable but why not everything? Of course once you know this, it's easy to set things as you want but it's just weird.
Overall in some ways the Radeon software for AMD GPUs is easier than AfterBurner for Nvidia GPUs but the numbers at least make sense in AB. Though tbh you also need to know that Nvidia GPUs downclock themselves as they warm up and throw off your manual voltage curves in AB, no nothing is perfect here. If you're manually stetting those GTX curves in AB, you need to wait until your GPU has cooled off and usually quit and restart AB to make sure you are setting everything in a consistent manner.
As far as I can tell, the cooler on this Pulse 5600XT is so overkill for the card that it doesn't downclock. I assume this is not the case for the 5700XT and may not be the case for midrange or cheaper 5700s. And that brings me to the main difference between these 2 cards and my experience with them:
They're about the same. And they're not.
In regular use they really are about the same, but their implementation is very different. At one time the 1080 was the fastest consumer card on the planet, so it's fully unlocked everywhere and if you have a good cooler and memory, you can squeeze a lot of extra performance from it. I don't have that cooler so my specific 1080 is hamstrung. However 1911 MHz and 12GBps capable memory do make for a very good gaming experience.
The 5600XT is specifically made to fit into a lower spot in a product stack so it's artificially limited in performance to have specific top clocks (~1775 MHz) so it doesn't compete with the 5700 (lol it does and beats it for value, thank you GTX 2060). So I can't OC it which is mildly annoying as the GPU clearly could if not software, but at least the memory can be OC'd and mine goes to about 14.7GHz dependably. And the Pulse 5600XT is simply a set it at max and forget it card as it would still be loafing around in a 30C room.
Frankly the cooling improvement is by far the best thing about the Pulse over my PNY and that is a very nice improvement. And it has one more advantage: I can now RMA my PNY to get the damn fan fixed.
Why did I do it? 'Cuz. Yeah, pretty much that's it. I had a lot of extra BezosBucks I hadn't originally planned to have which I was gonna spend on a ~$130 RX 570 and actually had enough for a decent current gen AMD card, the 5600XT.
The actual cards? I replaced a PNY GTX 1080 OC with a Sapphire Pulse 5600 XT.
The PNY is a bottom of the barrel 1080 with a good GPU and memory but a crud cooler design. The 1080 is a nominal 180W card and that can be extended 20% to 216W but the cooler is barely a 150W capable unit and frankly I'd call it a 120W unit. One of the 2 fans started rattling btw 65% and 80% fan speed after a few months so I set my manual fan curve in MSI Afterburner to hop directly from 65% to 80% at 74C. Lite & med gaming keeps below 74C as I manually undervolted the card (usually to 1822MHz and 0.85v) and if I went to 1911MHz and 0.90v (in Control or Jedi:FO), I'd usually set the fans to 80 or 85% manually.
All that and more to compensate for a crap cooler, though I got the card for $440 after RTX started shipping so it was a decent deal. And memory OCs to max 12GBps so there's great memory in this unit. I always intended it to be run undervolted and not madly overclocked as it was replacing a 1060 and even at 1822MHz is way faster. Frankly, if it hadn't been for the fan thing I'd be happy with it.
However I wanted to see how it is in Radeonville so I got the Sapphire Pulse 5600XT, which has been favorably reviewed everywhere so I knew I was getting probably the best deal on any midrange card available right now. Which is good as this is really just a fun buy.
Well, the Pulse's build is everything the PNY isn't, quite hefty with a taller cooler (still fits in my cheap case), and since the 5600XT is a purposefully limited unit so it doesn't compete with the 5700 too much (though it does and eliminates the 5700 as a decent buy), I have trouble getting it to run warmer than 62C. In fact on most games, it's rummaging around 56C and Control barely gets it to that 63C (in a ~26C room). I undervolt it to 0.918v at 1775MHz so it may run warmer at it's stock voltage which is somewhere around 0.98v.
It performs about the same in every game as the 1080 with no noticeable differences, though it does benchmark lower in Dirt Rally (about 120fps to the 1080's 135 fps or so @1440p) and it may be a frame or 2 faster in Control (both in the low 50s @1440p), but neither feel any different while playing. The biggest difference is in vanilla Minecraft of all things. The Pulse 5600XT gets about half the FPS of the 1080 though both are heavily bottlenecked by Minecraft's 3D implementation as both usually run at less than 20% utilization (with a i5-8400 which itself is only running at about 30% util). MC is crap like that. Both run better with OptiFine installed, at around 70FPS @1440p with long draw distances and mods.
So, the black screens or similar. I can make the 5600XT do a black screen easily and also avoid them entirely. Here's how: Afterburner. If I enable AB to have hardware-level access to the 5600XT, it outputs black screens on occasion during high utilization times (benchmarks, etc), though more rarely in-game. Don't give AB that hardware-level access and no black screens. Maybe there's another way to induce them using other software but I don't run anything else that does this. I can still use AB sans HW-level access for it's FPS counter with no black screens.
If you're coming to the Radeon side from the GTX/RTX side, and you're an enthusiast, you probably depend on AfterBurner to squeeze your GTX/RTX to it's limits and you simply need to let that go and use AMD's Radeon software to do the same. I have about 3 years squeezing blood from my 4 GTX cards with differing capabilities using AfterBurner so after 2 weeks in, I'm still in the learning phase and here's what I see:
Some of the numerical settings in the Radeon software do not correlate 1:1 with what TPU's GPU-Z reports or even the documentation for the card describe. WTF?? These are simply numbers and should be dead simple to get exactly correct. You set the card to 1820 MHz (max for the 5600XT) in Radeon SW and it tops out around 1775 MHz. All other speeds scale accordingly, it's not just a 45MHz buzzcut off the top. You set the fans to 35% and they run at 50% and also scale like this up and down. WTF? This is not rocket surgery. OK, the memory speed and voltages are correctly set and manageable but why not everything? Of course once you know this, it's easy to set things as you want but it's just weird.
Overall in some ways the Radeon software for AMD GPUs is easier than AfterBurner for Nvidia GPUs but the numbers at least make sense in AB. Though tbh you also need to know that Nvidia GPUs downclock themselves as they warm up and throw off your manual voltage curves in AB, no nothing is perfect here. If you're manually stetting those GTX curves in AB, you need to wait until your GPU has cooled off and usually quit and restart AB to make sure you are setting everything in a consistent manner.
As far as I can tell, the cooler on this Pulse 5600XT is so overkill for the card that it doesn't downclock. I assume this is not the case for the 5700XT and may not be the case for midrange or cheaper 5700s. And that brings me to the main difference between these 2 cards and my experience with them:
They're about the same. And they're not.
In regular use they really are about the same, but their implementation is very different. At one time the 1080 was the fastest consumer card on the planet, so it's fully unlocked everywhere and if you have a good cooler and memory, you can squeeze a lot of extra performance from it. I don't have that cooler so my specific 1080 is hamstrung. However 1911 MHz and 12GBps capable memory do make for a very good gaming experience.
The 5600XT is specifically made to fit into a lower spot in a product stack so it's artificially limited in performance to have specific top clocks (~1775 MHz) so it doesn't compete with the 5700 (lol it does and beats it for value, thank you GTX 2060). So I can't OC it which is mildly annoying as the GPU clearly could if not software, but at least the memory can be OC'd and mine goes to about 14.7GHz dependably. And the Pulse 5600XT is simply a set it at max and forget it card as it would still be loafing around in a 30C room.
Frankly the cooling improvement is by far the best thing about the Pulse over my PNY and that is a very nice improvement. And it has one more advantage: I can now RMA my PNY to get the damn fan fixed.
Similar threads
