The Effects of SMT and PBO on AMD Ryzen 9 7000X Processors

When it comes to multi-threaded performance, many gamers and enthusiasts are looking for ways to squeeze every last bit of power out of their processors. One way to do this is by enabling the Simultaneous Multithreading (SMT) feature in Windows 10. SMT allows each physical core on the processor to run two threads at the same time, effectively doubling the number of threads that can be utilized.

However, as AMD has discovered, enabling SMT comes with a cost. In lightly threaded situations, such as gaming, enabling SMT can actually sacrifice performance and introduce additional overhead. This is because the processor's idle resources are being used to create virtual processing threads, which can lead to increased power consumption and heat production. Additionally, some of the hardware resources are split in order to support this feature, which can result in potential slowdowns.

On the other hand, disabling SMT can free up some resources and potentially improve performance in lightly threaded situations. However, this comes at a cost - disabling SMT will always lead to less performance in multi-threaded situations. This is because the processor's idle resources are not being utilized as effectively, resulting in lower overall throughput.

In our testing, we discovered that disabling SMT on a stock Ryzen 9 7000X actually had a net positive effect on gaming performance. While the averages did get a slight bump, it was those Ultra-important 1% lows that saw the biggest impact. Simply turning off SMT had a bigger effect on gaming than throwing tons of power at the chip with Power Boost Overdrive (PBO).

However, there is a minimal effect when moving to a higher resolution, where CPU bottlenecking becomes less and more focus is put on GPU performance. In individual games, the results are actually highly variable - some titles see a pretty big overall jump in both averages and .1% lows, while others experience a small frame rate regression.

Our advice would be to test performance on whatever game you're playing at the very worst, keeping SMT on, and simply turning it off if possible. This way, you can enjoy a better gameplay experience without sacrificing too much potential performance. So, should you keep SMT enabled or disabled? The answer is up to you.

But what about PBO, you ask? Can it actually fix Zen 5 and the answer is yes and no. On one hand, you do get better performance - specifically, around 10% better performance in all-core workloads. However, this comes at a massive trade-off: an almost 80% increase in power consumption. This means that performance per watt gets completely thrown out of the window.

In our testing, we found that PBO gets about 10% better performance in all-core workloads, but this is largely due to the Ryzen 9 7000X having the most to gain from PBO, while other Zen 5 CPUs will actually see minimum impacts. This means that for those of us who are hoping for a straightforward optimization of the processor architecture, we're likely out of luck.

AMD's intent with Zen 5 is also worth noting. While some argue that they should be optimizing this architecture for the desktop market, it's very clear that their primary intent is to make Zen 5 the best possible performance in the server market. This has led to a trickle-down effect into the desktop market, which can result in AMD becoming a stronger company in the future.

As we wrap up our review of the Ryzen 9 7000X, it's clear that there's still much to be learned about these processors and their potential for performance optimization. While SMT and PBO offer some interesting possibilities, they also come with significant trade-offs. Whether you choose to enable or disable SMT, or whether you opt for PBO, the choice is ultimately yours.

"WEBVTTKind: captionsLanguage: enwell everybody by this point in time I'm pretty sure you've had your fill of ryzen 9000 Series reviews but in our initial video I left one question sort of unanswered and that is can Precision boost overdrive or PBO actually add enough performance to these chips in order for it to make any difference whatsoever and also since that initial review went live there's been a couple of other little tricks that have been published out there that I wanted to test and one of those is does smt actually add a little bit more gaming performance too so I know a lot of the reviews that you've seen have been I would say sort of like mixed and not very positive the same thing goes for hours but what I really wanted to see here is does zen 5 have any room left in the tank for additional performance well according to AMD just enabling PBO for additional performance should be a cakewalk since out of all of the Zen 5 chips the 9700 X I'm using here supposedly has the highest possible gains with PBO enabled and that sort of makes sense you see the ryzen 7 9700x comes with a rated TDP of just 65 Watts Which is far less than its predecessor the 7700x even in real world situations you can drive a truck into the Delta between their relative power consumption numbers that translates to cooling 2 so of course I got a lot of people thinking what happens when you end up running the 9700x at the exact same or very very similar power limits to the 7700x can Zen 5 really start to shine and you know what else really shines that's this AI from montech contact with another insane value product the hyperflow argb aiio available in black or white 240 or 360 sizes with full platform compatibility and friendly tools along the way that not only look gorgeous with pump lighting and matching blade illumination but also a performance that should satisfy with any hot silicon thanks to highspeed fans high density fins on the rod a good pump and daisy chain fans for easy Cable Management all backed by the six-year warranty so you don't have to worry Check It Out Below all right with all that out of the way how did we end up testing this look every single motherboard manufacturer approaches the secondary PBO settings differently some of them give you some options for temperature limits other ones have a a crazy amount of other fine grain options so for this what we wanted to do is simply level the playing field we enabled PBO and that's it that means that it maintains the chips 95° temperature ceiling while also giving it a little bit more room to flex its power and technically that should increase its frequencies well in gaming PBO allowed for an increase of about 13 Watts which really isn't much even in a game like cyberpunk which pushes the CPU much harder than most other titles as for temperatures we only saw moderate increases versus the stock 9700x even in a closed case as a matter of fact the 9700x is now running at almost exactly the same temperatures as the 7700x though its constant numbers are a bit lower so yeah it's still a bit easier to cool even with PBO enabled on a Peerless assassin and that additional power and corresponding heat does end up leading to a small but noticeable bump in constant frequencies I mean it's just under 90 MHz averaged across all cores but every little bit counts but where overdrive really hits hard is in an all core workload here our 9700x ended up hitting just over 140 Watts which is a far cry from its original 88 watts and almost identical to what the 7700x just guzzles back and all that power pumped into a single CCD leads to a stratospheric jump in heat so much so the Peerless assassin we're using was almost but not quite overwhelmed just like it was with the 7700x it comes very close to the Chip's maximum temperature too more importantly while the stock 9700x has a pretty big Delta between its peak and constant speeds that's pretty much evaporated here so frequency over time is a lot more stable than it was and that should provide a net positive for performance especially in some longer test that we normally run and at this point in time I absolutely know what everybody is going to be yelling at their screen about maybe the Peerless assassin running a 75% is actually holding back this chip because of the temperatures that we're seeing well let's add an Arctic liquid freezer 3 240 mm a to this and see what ends up happening and guess what temperatures didn't get lowered as much as you might expect that's likely due to amd's IHS is ability to conduct heat being the real limiting factor here rather than the cooler and those slightly lower temperatures allowed the chip to request a little bit bit more power to but overall we aren't looking at a night and day difference with an increase of just 3 wats on average simply put that's because this specific chip doesn't need the additional power to hit optimal frequencies even in longer workloads so the increase ends up being less than 20 MHz which basically amounts to a rounding error in these cases so that's how things shake out on the more technical side what we have now is the 9700x is taking just as much power as a 7700x and it's also just as hard to cool too but do all those sacrifices on the efficiency front actually lead to better performance that's what I wanted to find out and we're going to go through a couple of very very quick multi-core tests to see what happens and straight off the Jump PBO places the 9700 X right where a lot of people were actually hoping it would be in the first place that's a good 15% in front of the 7700x and 10% higher than its current 65 wat stock form but remember it's getting here while chugging back a ton more power the same thing goes for CCH 2024 with the 9700x suddenly showing a comfortable lead over the 7700x and in full core workloads like blender people will really be scratching their heads wondering why AMD just didn't launch this bloody thing with a higher power level to begin with because it's obviously got the chops to put on a much better show than it does at 65 watts and if these were out of the box results I'm sure we would have seen it costing more too and that's important to remember here because AMD they're not doing you really any favors by pricing this where it is right right now so we're sort of like in a lose lose situation but there's one area that won't be affected by PBO that's lightly threaded situations simply because they're not power limited in any way so the increase in input current thresholds doesn't have an impact whatsoever and that might end up just nuking potential in one area people might have been hoping for more performance and that's in gaming and that's really where our next stop is going to be 1080p 1440p gaming results and I'm sure everybody has some assumptions about where this is going to go and yet in some cases I was actually pretty shocked at the outcome because Precision boost overdrive actually benefited some games more than I thought it would at least at 1080p I mean there was never enough of a difference for me to suddenly jump up and start saying this is a vastly superior processor for gaming now but obviously PBO does have some sort of like Edge case benefits that are highly dependent on how game engines and amd's algorithms manage processing tasks between the cores in the vast majority of titles though performance is absolutely identical between a stock 9700x and one that has PBO enabled and that's likely one of the primary reasons why AMD decide to stick with a 65 wat TDP here there's just no tangible reason to run this chip at higher power levels in gaming either way I'm sure people who are looking for better multi-core performance wish they would have just bitten the bullet and launched this thing at a higher nominal power level the other thing I want to mention is that any benefits even those small frame rate increases are completely nullified when moving to a higher resolution like we're showing here with 1440p finally you'll also notice that PBO has very little to no impact on the .1% lows which is one area I know some people were hoping to see a bit more Improvement but that simply didn't happen on the positive side there weren't any performance regressions but this also points towards the one fact I was mentioning at the beginning of this video PBO won't do anything in situations where the CPU isn't power limited and most games don't push the 9700x anywhere close to its maximum power limit even when it's running in its default out of the box configuration and there's one other thing I wanted to discuss in this video and that is this discussion that's going on out there that potentially disabling smt or simultaneous multi-threading could lead to some performance benefits in gaming for these processors and a lot of that stems from the fact that there has been some core architectural changes to Zen 5 that could make disabling smt pretty beneficial at least on paper and look disabling smt is is nothing people have been doing this since Intel introduced hyperthreading way back in the the penium 4 days so they have been doing this in order to get a quick boost on one side but also it sacrifices multi-threaded performance on the other and the theory behind this well it's pretty simple smt uses the idle resources of each core to create a virtual processing thread while scaling isn't 100% perfect this allows Windows to recognize and utilize double the amount of cores or threads that are physically present within a given CPU but this approach also splits some ony Hardware resources which technically results in more overhead and introduces some potential slowdowns so basically two physical cores will always offer better performance than a single core producing two threads so like I said disabling smt will always lead to less performance in multi-threaded situations but it could technically also free up some resources in lightly threaded situations sort of like gaming and once you average things across all 16 games we test tested well I was pretty surprised by these results because based on the titles we're using disabling smt on a stock 9700x actually has a net positive effect and while the averages do get a bump it's those Ultra important. 1% lows that see the biggest impact as a matter of fact simply turning off smt has a bigger effect on gaming than throwing tons of power at the chip with PBO the other thing you need to take into account is there's a very minimal effect when moving to a higher resolution where CPU bottlenecking becomes less and more focus is put on GPU performance but the results in individual games are actually highly variable some like these get a pretty big overall jump in both averages and .1% lows to the point where flipping the smt switch feels like a no-brainer most of these are actually titles that are lightly threaded but still somewhat CPU limited on the other hand there are some situations that actually see a small frame rate regression it isn't anything dramatic but the last thing anyone wants is for performance to be left on the table so my advice to you would be to test performance on whatever game you're playing at the very worst you keep smt on and in the best case scenario you simply turn it off and enjoy a better gameplay experience so let's get back to PBO for a second the question here was can it actually fix Zen 5 and the answer to that is yes and no on one hand you do get better performance performance that a lot of people were expecting Straight Out of the Box on the other hand it comes with a massive massive trade-off when it comes to power consumption and heat production and that goes to show that performance rarely SC sces in linear fashion when poers increased Beyond a certain point there's a lot of diminishing returns here and with the 9700x it feels like AMD hit their best possible blend of efficiency and throughput and yes I know some people simply don't appreciate that in this case PBO gets you about 10% better performance in all core workloads right alongside an almost 80% increase I'm going to repeat that again an 80% increase in power consumption basically its performance per watt gets completely thrown out of the window I mean it's good to know that uplifts are there at the flip of a bio switch if you want it and if your cooler can handle it but there's literally no positive impacts for gaming and to make matters even worse this is a best case scenario since according to AMD themselves the 9700x has the most to gain from PBO while all other Zen 5 CPUs will actually see minimum impacts anyways to finish this video I wanted to take you down sort of like a personal Journey that I had during this whole review process and also this video video the team and I we were talking the other day and we basically came to the conclusion that we as well as a lot of other people out there are simply grasping at straws when it comes to Zen 5 we're hoping that someway somehow we can fiddle with the settings and make these processors what we had hoped they would be straight out of the box and I think a lot of that comes down to amd's intent with Zen 5 it's very very obvious as you go through this that they are not trying to optimize this architecture necessarily for the desktop Market that is not their primary intent their primary intent to make Zen 5 probably the best it can be in the server Market that is simply where the money is we get some of the trickle down effects in the desktop Market whether you like that or you don't or that potentially leads to AMD becoming a sort of like a stronger company in the future well we're just going to have to see and until that point well I'm Mike with har conu I hope that you enjoyed this video it went a little bit deeper than we normally would into some additional testing and I'm going to see you in the next one have a great day guyswell everybody by this point in time I'm pretty sure you've had your fill of ryzen 9000 Series reviews but in our initial video I left one question sort of unanswered and that is can Precision boost overdrive or PBO actually add enough performance to these chips in order for it to make any difference whatsoever and also since that initial review went live there's been a couple of other little tricks that have been published out there that I wanted to test and one of those is does smt actually add a little bit more gaming performance too so I know a lot of the reviews that you've seen have been I would say sort of like mixed and not very positive the same thing goes for hours but what I really wanted to see here is does zen 5 have any room left in the tank for additional performance well according to AMD just enabling PBO for additional performance should be a cakewalk since out of all of the Zen 5 chips the 9700 X I'm using here supposedly has the highest possible gains with PBO enabled and that sort of makes sense you see the ryzen 7 9700x comes with a rated TDP of just 65 Watts Which is far less than its predecessor the 7700x even in real world situations you can drive a truck into the Delta between their relative power consumption numbers that translates to cooling 2 so of course I got a lot of people thinking what happens when you end up running the 9700x at the exact same or very very similar power limits to the 7700x can Zen 5 really start to shine and you know what else really shines that's this AI from montech contact with another insane value product the hyperflow argb aiio available in black or white 240 or 360 sizes with full platform compatibility and friendly tools along the way that not only look gorgeous with pump lighting and matching blade illumination but also a performance that should satisfy with any hot silicon thanks to highspeed fans high density fins on the rod a good pump and daisy chain fans for easy Cable Management all backed by the six-year warranty so you don't have to worry Check It Out Below all right with all that out of the way how did we end up testing this look every single motherboard manufacturer approaches the secondary PBO settings differently some of them give you some options for temperature limits other ones have a a crazy amount of other fine grain options so for this what we wanted to do is simply level the playing field we enabled PBO and that's it that means that it maintains the chips 95° temperature ceiling while also giving it a little bit more room to flex its power and technically that should increase its frequencies well in gaming PBO allowed for an increase of about 13 Watts which really isn't much even in a game like cyberpunk which pushes the CPU much harder than most other titles as for temperatures we only saw moderate increases versus the stock 9700x even in a closed case as a matter of fact the 9700x is now running at almost exactly the same temperatures as the 7700x though its constant numbers are a bit lower so yeah it's still a bit easier to cool even with PBO enabled on a Peerless assassin and that additional power and corresponding heat does end up leading to a small but noticeable bump in constant frequencies I mean it's just under 90 MHz averaged across all cores but every little bit counts but where overdrive really hits hard is in an all core workload here our 9700x ended up hitting just over 140 Watts which is a far cry from its original 88 watts and almost identical to what the 7700x just guzzles back and all that power pumped into a single CCD leads to a stratospheric jump in heat so much so the Peerless assassin we're using was almost but not quite overwhelmed just like it was with the 7700x it comes very close to the Chip's maximum temperature too more importantly while the stock 9700x has a pretty big Delta between its peak and constant speeds that's pretty much evaporated here so frequency over time is a lot more stable than it was and that should provide a net positive for performance especially in some longer test that we normally run and at this point in time I absolutely know what everybody is going to be yelling at their screen about maybe the Peerless assassin running a 75% is actually holding back this chip because of the temperatures that we're seeing well let's add an Arctic liquid freezer 3 240 mm a to this and see what ends up happening and guess what temperatures didn't get lowered as much as you might expect that's likely due to amd's IHS is ability to conduct heat being the real limiting factor here rather than the cooler and those slightly lower temperatures allowed the chip to request a little bit bit more power to but overall we aren't looking at a night and day difference with an increase of just 3 wats on average simply put that's because this specific chip doesn't need the additional power to hit optimal frequencies even in longer workloads so the increase ends up being less than 20 MHz which basically amounts to a rounding error in these cases so that's how things shake out on the more technical side what we have now is the 9700x is taking just as much power as a 7700x and it's also just as hard to cool too but do all those sacrifices on the efficiency front actually lead to better performance that's what I wanted to find out and we're going to go through a couple of very very quick multi-core tests to see what happens and straight off the Jump PBO places the 9700 X right where a lot of people were actually hoping it would be in the first place that's a good 15% in front of the 7700x and 10% higher than its current 65 wat stock form but remember it's getting here while chugging back a ton more power the same thing goes for CCH 2024 with the 9700x suddenly showing a comfortable lead over the 7700x and in full core workloads like blender people will really be scratching their heads wondering why AMD just didn't launch this bloody thing with a higher power level to begin with because it's obviously got the chops to put on a much better show than it does at 65 watts and if these were out of the box results I'm sure we would have seen it costing more too and that's important to remember here because AMD they're not doing you really any favors by pricing this where it is right right now so we're sort of like in a lose lose situation but there's one area that won't be affected by PBO that's lightly threaded situations simply because they're not power limited in any way so the increase in input current thresholds doesn't have an impact whatsoever and that might end up just nuking potential in one area people might have been hoping for more performance and that's in gaming and that's really where our next stop is going to be 1080p 1440p gaming results and I'm sure everybody has some assumptions about where this is going to go and yet in some cases I was actually pretty shocked at the outcome because Precision boost overdrive actually benefited some games more than I thought it would at least at 1080p I mean there was never enough of a difference for me to suddenly jump up and start saying this is a vastly superior processor for gaming now but obviously PBO does have some sort of like Edge case benefits that are highly dependent on how game engines and amd's algorithms manage processing tasks between the cores in the vast majority of titles though performance is absolutely identical between a stock 9700x and one that has PBO enabled and that's likely one of the primary reasons why AMD decide to stick with a 65 wat TDP here there's just no tangible reason to run this chip at higher power levels in gaming either way I'm sure people who are looking for better multi-core performance wish they would have just bitten the bullet and launched this thing at a higher nominal power level the other thing I want to mention is that any benefits even those small frame rate increases are completely nullified when moving to a higher resolution like we're showing here with 1440p finally you'll also notice that PBO has very little to no impact on the .1% lows which is one area I know some people were hoping to see a bit more Improvement but that simply didn't happen on the positive side there weren't any performance regressions but this also points towards the one fact I was mentioning at the beginning of this video PBO won't do anything in situations where the CPU isn't power limited and most games don't push the 9700x anywhere close to its maximum power limit even when it's running in its default out of the box configuration and there's one other thing I wanted to discuss in this video and that is this discussion that's going on out there that potentially disabling smt or simultaneous multi-threading could lead to some performance benefits in gaming for these processors and a lot of that stems from the fact that there has been some core architectural changes to Zen 5 that could make disabling smt pretty beneficial at least on paper and look disabling smt is is nothing people have been doing this since Intel introduced hyperthreading way back in the the penium 4 days so they have been doing this in order to get a quick boost on one side but also it sacrifices multi-threaded performance on the other and the theory behind this well it's pretty simple smt uses the idle resources of each core to create a virtual processing thread while scaling isn't 100% perfect this allows Windows to recognize and utilize double the amount of cores or threads that are physically present within a given CPU but this approach also splits some ony Hardware resources which technically results in more overhead and introduces some potential slowdowns so basically two physical cores will always offer better performance than a single core producing two threads so like I said disabling smt will always lead to less performance in multi-threaded situations but it could technically also free up some resources in lightly threaded situations sort of like gaming and once you average things across all 16 games we test tested well I was pretty surprised by these results because based on the titles we're using disabling smt on a stock 9700x actually has a net positive effect and while the averages do get a bump it's those Ultra important. 1% lows that see the biggest impact as a matter of fact simply turning off smt has a bigger effect on gaming than throwing tons of power at the chip with PBO the other thing you need to take into account is there's a very minimal effect when moving to a higher resolution where CPU bottlenecking becomes less and more focus is put on GPU performance but the results in individual games are actually highly variable some like these get a pretty big overall jump in both averages and .1% lows to the point where flipping the smt switch feels like a no-brainer most of these are actually titles that are lightly threaded but still somewhat CPU limited on the other hand there are some situations that actually see a small frame rate regression it isn't anything dramatic but the last thing anyone wants is for performance to be left on the table so my advice to you would be to test performance on whatever game you're playing at the very worst you keep smt on and in the best case scenario you simply turn it off and enjoy a better gameplay experience so let's get back to PBO for a second the question here was can it actually fix Zen 5 and the answer to that is yes and no on one hand you do get better performance performance that a lot of people were expecting Straight Out of the Box on the other hand it comes with a massive massive trade-off when it comes to power consumption and heat production and that goes to show that performance rarely SC sces in linear fashion when poers increased Beyond a certain point there's a lot of diminishing returns here and with the 9700x it feels like AMD hit their best possible blend of efficiency and throughput and yes I know some people simply don't appreciate that in this case PBO gets you about 10% better performance in all core workloads right alongside an almost 80% increase I'm going to repeat that again an 80% increase in power consumption basically its performance per watt gets completely thrown out of the window I mean it's good to know that uplifts are there at the flip of a bio switch if you want it and if your cooler can handle it but there's literally no positive impacts for gaming and to make matters even worse this is a best case scenario since according to AMD themselves the 9700x has the most to gain from PBO while all other Zen 5 CPUs will actually see minimum impacts anyways to finish this video I wanted to take you down sort of like a personal Journey that I had during this whole review process and also this video video the team and I we were talking the other day and we basically came to the conclusion that we as well as a lot of other people out there are simply grasping at straws when it comes to Zen 5 we're hoping that someway somehow we can fiddle with the settings and make these processors what we had hoped they would be straight out of the box and I think a lot of that comes down to amd's intent with Zen 5 it's very very obvious as you go through this that they are not trying to optimize this architecture necessarily for the desktop Market that is not their primary intent their primary intent to make Zen 5 probably the best it can be in the server Market that is simply where the money is we get some of the trickle down effects in the desktop Market whether you like that or you don't or that potentially leads to AMD becoming a sort of like a stronger company in the future well we're just going to have to see and until that point well I'm Mike with har conu I hope that you enjoyed this video it went a little bit deeper than we normally would into some additional testing and I'm going to see you in the next one have a great day guys\n"