AMD 7900X Retested – PBO, SMT & Overclocked

Last week, we brought you a review of the Ryzen 7 9700X, the latest release in a seemingly long line of rather boring new CPU launches that offer minor upgrades to existing products. In short, it's fair to say we weren't particularly impressed by the small increase in performance it offered over a range of benchmarks, but more so with the gaming performance. We went back and retested with PBO, as we were called out for not enabling it. However, since PBO also works on the CPUs it replaced, the performance gap remained basically the same. So, here we are again, trying to milk the views, I mean, find the performance that we all feel should be there. Surely there's some magic bullet in the settings that will make the 9700X shine brighter and make the upgrade worth it!? Well according to a recent feature on TechPowerUp, the secret lies in disabling SMT so we thought we’d go about testing it for ourselves. Historically SMT being disabled has improved gaming performance in some games and decreased it in others, so that's what we're going to try today. Simultaneous multithreading, or SMT, is a process that allows a CPU core to operate multiple jobs at once, and while this does have performance benefits for most cases, disabling it is sort of like forcing the CPU to focus more specifically on a given task. But are users expected to enable SMT for productivity then disable it when they want to game, then put it back on when they’re done? That’s a bit of a clunky process, and one that requires a system reboot too, so it’s not likely they will. We're also under no illusion that the 9700X has a lower TDP so it's not all about performance, but we are, so we want to try and get the most out of it. We've also re-tested with PBO set to max limits, and the same but with SMT disabled too to give a much broader spectrum of results. Honestly, if the performance isn’t here, it isn’t anywhere, and we’re not going to shine coal into a diamond with any more convoluted BIOS settings though I am currently working on a 5.3GHz all core overclock, so maybe I’ll make some content on that soon. With all that in mind though, let’s get to the benchmarks, as I’m sure you’re eager to see the results! How We Test We continue to update our testing methods around once per year. As such, we re-test older hardware to reflect changes over time. These can be driver updates, Windows updates, game patches, and more; all of which have an impact on performance figures. Furthermore, we update our test benches to newer and more relevant hardware. This means that our new reviews aren’t always comparable to those of older reviews, so please compare the testing methodology on older reviews should you be trying to compare them with newer ones. Replicate Our Results When it comes to our benchmarks in our reviews, the benchmarks are pretty self-explanatory and are kept as simple as possible, although there are a few exceptions. Remember that your choice of a graphics card, CPU, the silicon lottery, and other factors can yield different numbers, and there’s always a margin for error when using any software. Links are provided below, as well as the settings we use. We encourage you to not just look at how one product compares to any other, but how it compares to your own. If you’re looking to build a new system, you should benchmark your current PC using the benchmarks available to you. You should then look at the percentage improvement from your current hardware to the hardware tested here to give you a ballpark figure of how much of an upgrade this will provide you with. Processor Test Systems Test System (AM5) Motherboard Gigabyte X670E AORUS Master Review Buy Now Processor Changes Per Review Review Buy Now GPU INNO3D RTX 4090 iCHILL X3 OC Review Buy Now RAM TEAMGroup T-Force Delta RGB 7600MHz CL36 Review Buy Now CPU Cooler NZXT Kraken Z73 RGB Buy Now Power Supply NZXT C1000 Gold Review Buy Now Storage Seagate FIRECUDA 530 1TB Review Buy Now Chassis NZXT H7 Flow Review Buy Now OS Windows 11 Pro Version 21H2 Drivers Graphics: 537.42 Test System (AM4) Motherboard Gigabyte X570S AORUS Master Review Buy Now Processor Changes Per Review Review Buy Now GPU INNO3D RTX 4090 iCHILL X3 OC Review Buy Now RAM Kingston Fury Renegade DDR4 32GB 3600MHz CL16 Review Buy Now CPU Cooler NZXT Kraken Z73 RGB Buy Now Power Supply NZXT C1000 Gold Review Buy Now Storage Seagate FIRECUDA 530 1TB Review Buy Now Chassis NZXT H7 Flow Review Buy Now OS Windows 11 Pro Version 21H2 Drivers Graphics: 537.42 Test System (Z690) Motherboard MSI Z690 UnifyBIOS Version 7D28v17 Review Buy Now Processor Changes Per Review GPU INNO3D RTX 4090 iCHILL X3 OC Review Buy Now RAM TEAMGroup T-Force Delta RGB 7600MHz CL36 Review Buy Now CPU Cooler NZXT Kraken Z73 RGB Buy Now Power Supply NZXT C1000 Gold Review Buy Now Storage Seagate FIRECUDA 530 1TB Review Buy Now Chassis NZXT H7 Flow Review Buy Now OS Windows 11 Pro Version 21H2 Drivers Graphics: 537.42 Test System (Z790) Motherboard ASUS ROG Maximus Z790 Extreme Review Buy Now Processor Changes Per Review GPU INNO3D RTX 4090 iCHILL X3 OC Review Buy Now RAM TEAMGroup T-Force Delta RGB 7600MHz CL36 Review Buy Now CPU Cooler NZXT Kraken Z73 RGB Buy Now Power Supply NZXT C1000 Gold Review Buy Now Storage Seagate FIRECUDA 530 1TB Review Buy Now Chassis NZXT H7 Flow Review Buy Now OS Windows 11 Pro Version 21H2 Drivers Graphics: 537.42 Software We use a variety of software from different developers to assist in our testing and to give us the most consistent results possible. Some applications are specifically used for testing, while others are simply used for monitoring and to aid us in making sure that our results are as accurate as possible. Software Used 3DMark FireStrike & TimeSpy (download) AIDA64 (download) Blender 3.6.0 Classroom (download) CapFrameX 1.7.0 (download) Cinebench R23 (download) Corona 1.3 (download) CPU-Z (download) Geekbench 5 (download) Google Octane 2.0 (link) GPU-Z (download) HWMonitor (download) Keyshot Viewer 11.2.1 (download) Mozilla Kraken 1.1 (link) PCMark 10 Express (download) PugetBench For Davinci Resolve Super PI 2.1 WP (download) V-Ray 5.00.20 (download) WebXPRT 4 (link) Gaming To gauge performance levels which are easily reproduced time after time, by both ourselves and our readers, we now stick strictly to games default profiles. Given how powerful most modern cards are getting, we've began using more demanding profiles for most games, but will otherwise state on a per game basis if we use other profiles beyond High, Ultra, Maximum or their respective equivalent at all resolutions. V-Sync is always disabled in our testing, and all FPS caps that can be removed are. We also enable XMP and Resizable bar technology across all platforms to keep things as fair as possible. Games While we test a lot of games, it's not possible to display all of the data on the website or in our videos. Instead, we include access to all of our testing data for our Patreon members (link), where members get a ton of other benefits as well. What we do, however, is display charts for averages across all game titles to give a better picture of performance across all titles tested. A Plague Tale Requiem Cyberpunk 2077 F1 23 Far Cry 6 Hogwarts Legacy Marvel's Spider-Man Remastered Microsoft Flight Simulator 2020 Ratchet & Clank: Rift Apart Remnant II Shadow of the Tomb Raider Starfield Resolutions 1080p (Full HD) Some tests will also include features such as NVIDIA DLSS and Ray Tracing. However, these will be clearly labelled and may only appear in reviews for graphics cards that support those features. Power Consumption Not being content on simply reading total system power draw at both idle and load, we now have the ability thanks to sensors built into the system to find the exact CPU package power draw during any one instance. We now check power draw during Cinebench R23 single and multi-core runs as well as during a set amount of games. Temperatures Temperatures are measured in a very similar way to power consumption and are taken during a Cinebench R23 multi-core and single-core run, as well as during a set amount of games at the max levels seen. Starting things off with 3DMark Timespy and disabling SMT has quite a detrimental effect to the tune of 12%. Enabling PBO on max limits does push performance back up by 2% but for the best performance, at least in this test, we’d advise to just straight up enable PBO with no other changes. Moving onto Blender and it’s a similar story, though expected, that the 9700X performs worse with SMT off as it sees a hit of 26% and enabling PBO on max limits further drops performance by 9% so again, the best you can expect with the 9700X is to straight up enable PBO and not touch anything else beyond going further with a curve optimiser that could net some further but still small gains. In Corona, it’s the same again. This time we see a 55% increase in render time by disabling SMT and PBO Max along with SMT disabled sees things drop another 12%. If you hadn’t noticed the trend so far, the 9700X with PBO is the best outcome at 66 seconds to render, but SMT could help more in terms of games that we’ll look at shortly. In V-Ray, we see another large drop in performance with 29% less Vsamples compared to the stock 9700X result and combining SMT disabled with PBO on max limits, that drops by 6% while PBO gives our best result at 17986, which is very similar to PBO on max limits, and a nice uplift from the stock score of 15539. Cinebench originally came in with a multi-core score of 19804 and PBO and PBO max saw this increase and with SMT disabled, as we’re talking multi-core, that score dropped by 24% , though on this occasion, PBO at max limits did see it recover, but only by 5%. The single-core score didn’t move much with just a margin of error 1% between the stock result and SMT result, but the best balance of multi-core and single-core scores come in when PBO was set to its max limits. Our last synthetic benchmark is Geekbench and here, again we see huge drops in performance in terms of the multi-threaded score with the 9700X with PBO enabled coming in with the best performance, while SMT drops performance at quite a hefty level, though single threaded performance again, stays within the same range as the PBO score, but native stock settings again seem to be the best for single-thread performance. A Plague Tale Follow the grim tale of young Amicia and her little brother Hugo, in a heartrending journey through the darkest hours of history. Hunted by Inquisition soldiers and surrounded by unstoppable swarms of rats, Amicia and Hugo will come to know and trust each other. As they struggle to survive against overwhelming odds, they will fight to find purpose in this brutal, unforgiving world. Available now on Steam. So now the big one, gaming and starting things off with A Plague Tale and in this game, we really are limited across the board across all of our CPUs. It’s a game that we’ve used in all CPU tests, but may relook at the settings or scene that we use to try and differentiate our results, but we can clearly see that any change to the 9700X makes no difference outside of margin of error. Cyberpunk 2077 Cyberpunk 2077 is an open-world, action-adventure RPG set in the megalopolis of Night City, where you play as a cyberpunk mercenary wrapped up in a do-or-die fight for survival. Improved and featuring all-new free additional content, customize your character and playstyle as you take on jobs, build a reputation, and unlock upgrades. The relationships you forge and the choices you make will shape the story and the world around you. Legends are made here. What will yours be? Available now on Steam. In Cyberpunk, our 9700X originally came in at 99 FPS and it’s here where SMT does…absolutely nothing apart from see our 1% low drop by a pretty staggering 10%. Luckily, enabling PBO on max limits does bring things back up again and beyond the stock result, as it now comes in 5% faster though PBO with max limits is still the winner as that packs another 7% performance over the stock results at 106 FPS and with a slightly stronger, but still small 1% low. Enabling Ray Tracing, naturally sees performance drop but what’s interesting here is how PBO matches with the stock result, while SMT, PBO Max or a combination of the two see performance drop and in some cases by 14%, so I guess disabling SMT isn’t looking to be this huge saving grace that we thought it may be. F1 23 Be the last to brake in EA SPORTS™ F1® 23, the official video game of the 2023 FIA Formula One World Championship™. A new chapter in the thrilling "Braking Point" story mode delivers high-speed drama and heated rivalries. Available now on Steam. In F1 23, we do see an 8 FPS gain when disabling SMT, but at over 300 frames per second, that equates to just 2% and comes in nigh on identical to the PBO and PBO Max figures, so for the sake of the gains you see, it doesn’t seem like the juice is worth the squeeze, at least in this title. Hogwarts Legacy Hogwarts Legacy is an open-world action RPG set in the world first introduced in the Harry Potter books. Embark on a journey through familiar and new locations as you explore and discover magical beasts, customize your character and craft potions, master spell casting, upgrade talents and become the wizard you want to be. Experience Hogwarts in the 1800s. Your character is a student who holds the key to an ancient secret that threatens to tear the wizarding world apart. Make allies, battle Dark wizards, and ultimately decide the fate of the wizarding world. Your legacy is what you make of it. Live the Unwritten. Available now on Steam. In Hogwarts, it’s another case of SMT being disabled not doing anything of any substance as we see a single FPS variance to the default result and PBO Max this time comes out on top, though we’re pretty high in terms of performance anyway, so any small difference wouldn’t be of notice to the gamer playing. Marvel's Spider-Man Remastered In Marvel’s Spider-Man Remastered, the worlds of Peter Parker and Spider-Man collide in an original, action-packed story. Play as an experienced Peter Parker, fighting big crime and iconic villains in Marvel’s New York. Web-swing through vibrant neighbourhoods and defeat villains with epic takedowns. Available now on Steam. Booting up Spider-Man and we finally see a game that’s able to utilise SMT being disabled as we now see an 11% increase in frames over the stock 9700X result and enabling PBO at max limits in conjunction with SMT being off increases that performance, albeit by a 1% margin, but it does mean that the combination pushes out 13% more performance over the stock result of 184 FPS. Enabling Ray Tracing and we see a similar trend with SMT being disabled gives us a 3% uplift, and combining that with PBO at max limits sees that increase again, but we’re then at a point of margin of error. PBO set to Max and keeping SMT is the best result here for the new Zen 5 CPU at 124 FPS, so it seems the CPU knows better at automatically adjusting that potentially what the user does. Microsoft Flight Simulator 2020 The Microsoft Flight Simulator Game of the Year (GOTY) Edition celebrates all the positive feedback, reviews and awards garnered, but it is also a “Thank You” to our fans – new and old – that have made Microsoft Flight Simulator such a phenomenon. Best of all, the Microsoft Flight Simulator Game of the Year Edition will be available as a free update for existing players on PC. Available now on Steam. Flight Sim is a very CPU intensive game, but in all honesty, the 9700X just runs out of steam and nothing we do to push the performance is of any significant importance. At most we’re talking a couple of frames per second with SMT off and PBO maxed out, but even then we’re talking margin of error and re-testing could garner the same or at least similar results again. Ratchet & Clank: Rift Apart With all new graphical and technical enhancements for deeper immersion, Red Dead Redemption 2 for PC takes full advantage of the power of the PC to bring every corner of this massive, rich and detailed world to life including increased draw distances; higher quality global illumination and ambient occlusion for improved day and night lighting; improved reflections and deeper, higher resolution shadows at all distances; tessellated tree textures and improved grass and fur textures for added realism in every plant and animal. Available now on Steam. In Ratchet & Clank, again, performance is very close between all of the various scenarios of PBO, PBO Max, SMT off, combined with PBO Max and the stock result too so there’s nothing really much to say other than performance doesn’t really change no matter the situation. Then when we enable Ray Tracing, the numbers are different but it’s a very similar affair with a small uplift with SMT off and PBO set to max but we’re talking minor numbers that can be deemed as margin of error, so it’s nothing to jump off your seat about, as it equals less than 1%. Remnant II Remnant II is the sequel to the best-selling game Remnant: From the Ashes that pits survivors of humanity against new deadly creatures and god-like bosses across terrifying worlds. Play solo or co-op with two other friends to explore the depths of the unknown to stop an evil from destroying reality itself. To succeed, players will need to rely on their own skills and those of their team to overcome the toughest challenges and to stave off humanity’s extinction. Available now on Steam. Remnant II is another game that is very CPU intensive, and due to that, we do start to see some sizable uplifts in performance. The 9700X came in originally at 151 FPS at stock defaults while disabling SMT saw that increase by 11%, now at 168, though that’s where our performance gains end as combining it with PBO on max limits saw the result come in identically in both the averages and the 1% lows. Shadow of the Tomb Raider Experience Lara Croft’s defining moment as she becomes the Tomb Raider. In Shadow of the Tomb Raider, Lara must master a deadly jungle, overcome terrifying tombs, and persevere through her darkest hour. As she races to save the world from a Maya apocalypse, Lara will ultimately be forged into the Tomb Raider she is destined to be. Available now on Steam. Shadow Of The Tomb Raider saw a small margin of error increase with SMT disabled, but it’s the combination with PBO set to max that saw the bigger gain of 5%, now at 243 FPS, but this matches the PBO Max result without SMT being disabled, so it’s clear to see that it’s PBO giving us the gain, and not us disabling SMT. Starfield Starfield is the first new universe in over 25 years from Bethesda Game Studios, the award-winning creators of The Elder Scrolls V: Skyrim and Fallout 4. In this next-generation role-playing game set amongst the stars, create any character you want and explore with unparalleled freedom as you embark on an epic journey to answer humanity’s greatest mystery. Available now on Steam. Lastly, in Starfield, we again see uplifts coming thanks to Precision Boost Overdrive, though we’re only talking up to 3%, while SMT has the opposite effect and sees performance tank by a pretty staggering 18%, with PBO Max being the only thing that raises performance back up again but only by a small margin, and SMT on this title should be avoided at all costs. Going One Step Further So looking at individual data, let’s break it down into two sections. The first being synthetics, which in anything that utilises multi-threaded utilization, SMT being disabled showed some really big dips in performance, but nothing beyond what’s expected when you’re literally telling the CPU to stop utilising its ability to run multiple threads per core and will instead only run a single thread per core, so with the 9700X which has 8 cores and 16 threads, the machine now sees it as only having 8 threads. In gaming though, where most games are geared towards a smaller number of cores of threads, you’d expect, depending on the game to see an uplift in performance, if its able to put more focus on the relevant cores, and sadly, when it comes to performance, we’ve seen some good and some bad, but when you average it out, the impact is actually pretty underwhelming. The AMD Ryzen 9700X managed an average FPS of 160, which is just 7 FPS above the performance of the 7700X, so not exactly a huge leap, but that’s just the base performance. When we enabled PBO, the 9700X increased its performance to 162, again, a very small improvement of just 2 FPS, and not really much to talk about there. So we tried it with SMT turned off, and that saw the performance increase to….163, again, that’s pretty much within a margin for error, and nothing you would really notice in real world usage. What’s even more surprising is that with SMT off and PBO set to max, the performance is back at the original 160 FPS average. That’s right, turning on all of the performance boosting technologies gave us the square root of nothing when we average out the gaming performance and due to that the cost per frame does move around, but we’re talking a cent or two here and there kind of making this whole experiment mathematically pointless, unless you play Spider-Man and want that big gain, because that really is the main game, I guess along with Remnant II that saw some fairly healthy uplifts. Sadly, games like Starfield brought that back down, hence the results being so close. SMT gave with one hand, and took away with the other. The big stark result on this chart though, is the 7800X3D which is just sitting there obnoxiously faster than everything else, and did that require a lot of extra trickery to get the performance to work? Nope. Final Thoughts Now don’t get me wrong. I’m more than aware that the 7800X3D is a CPU aimed at gaming, and the 9700X isn’t. It’s more of an all-rounder, but it’s not my fault that the 9700X and 7800X3D are so similar in price, so if you were eyeing it up, of course I’m going to be honest and tell you to pay $16 more and to grab it because on average, it is 14% faster and comes in cheaper in terms of the cost per frame. If you don’t want to spend that much, then just get the 7700X which with PBO, even at default levels isn’t too far away from the 9700X, and will save you $70. I set out to test this because I honestly thought, mainly based on comments from our viewers that I must have missed something with the 9700X. That’s why we did a re-review and tested with PBO, but then took it one step further with disabling SMT and then because I didn’t want to have to go through the pain of talking about PBO again, we combined them, and even that wasn’t enough to do things justice. At this point, my original verdict still stands and even though we don’t have any other Zen 5 parts right now at the time of filming, I’m pretty sure that they will follow the same trend and be lacklustre in the same fashion, at least in terms of performance. Sure, we can expect, much like with the 9700X that efficiency is better, but putting my mind into that of the consumer, I don’t think consumers, and mainly gamers care about efficiency as much as what AMD think they do. They want raw performance, and a generational uplift and while AMD promised an IPC improvement, that doesn’t translate natively to games, and that’s where AMD’s biggest consumer marketbase is. Going into this, and the reason we have made quite a bit of content on the 9700X is that I do feel for gamers who were looking at buying Zen 5, and mainly the 9700X. It’s not the best move AMD has done, and in a time where Intel are having their own fair share of bad press, this could have been a pivotal time for the red team, and sadly, that’s just not happened and I can’t see a way that AMD can redeem Zen 5, short of severely dropping the price, or somehow unlocking more performance. Luckily, we’re going to also put that to the test soon as we’ve overclocked the 9700X to 5.3GHz on all cores, so make sure you stick around for that one coming up soon.