AMD Ryzen Boost Frequency Fix Benchmarked: AGESA ABBA Tested
video description
Date: 2020-05-06
Related videos
Comments and reviews: 10
SomethingDarkseid
My PC is a fully functioning workstation and gaming PC and has things running in the background, such as iCue, Malwarebytes, so you instead of it just being a fresh install of Windows, people can see how this compares to GN testing, although I'm obviously not as methodical in my testing, but the results are pretty much the same when compared to GN On the Aorus Elite x570 with a 3900x and between the F4 and F5a and experiencing pretty much the same on a multi core workload, the same as Gamers Nexus. With the F5a only barely getting ahead but really, not by anything that would be considered significant with the clock speed ranging from 3. 9 GHz to 4. 1 GHz with heavy voltage fluctuations but the temperatures were about 5c lower on average. Temperatures never touched 70c and averaged around 68c in a 24c room whereas before, it would be hitting 73c on average in the same ambient. Single threaded performance saw an increase of approximately 5% over 4 runs on Cinebench R15 with a max turbo of 4. 5 GHz which I have never seen before on this CPU. This is within line from what GN has also experienced if you take into account that each CPU will act differently and also this is not a clean OS installation and has been running for over 2 months, since getting the 3900x and is my daily driver. The Cinebench R15 multi threaded score is an average of 3002 over 4 tests. This score isn't great, but with real world workloads, the 200 points is pretty much meaningless if not then it's pretty damn close to being meaningless. The Cinebench R15 single threaded scored 199. I am also running my RAM overclocked to 3600 MHz with 16. 19. 19. 19. 38 timings. I can't do the timings any tighter, as they'll just crash the system and even freeze the BIOS (that was fun) This BIOS update has pretty much cured my itch of wanting to overclock. Sorry for the long comment, hopefully this also helps, someone although again, no where near as methodical as GN but the end results are similar with the Cinebench R15 scores and real world workloads. Build 3900x X570 Gigabyte Aorus Elite GTX 1080 Ti 2x 16 GB DDR4 3200 OC'd to 3600 On a custom loop. Thermal paste is MX 2. My Grizzly Kryonaut arrived a bit late.
reply
My PC is a fully functioning workstation and gaming PC and has things running in the background, such as iCue, Malwarebytes, so you instead of it just being a fresh install of Windows, people can see how this compares to GN testing, although I'm obviously not as methodical in my testing, but the results are pretty much the same when compared to GN On the Aorus Elite x570 with a 3900x and between the F4 and F5a and experiencing pretty much the same on a multi core workload, the same as Gamers Nexus. With the F5a only barely getting ahead but really, not by anything that would be considered significant with the clock speed ranging from 3. 9 GHz to 4. 1 GHz with heavy voltage fluctuations but the temperatures were about 5c lower on average. Temperatures never touched 70c and averaged around 68c in a 24c room whereas before, it would be hitting 73c on average in the same ambient. Single threaded performance saw an increase of approximately 5% over 4 runs on Cinebench R15 with a max turbo of 4. 5 GHz which I have never seen before on this CPU. This is within line from what GN has also experienced if you take into account that each CPU will act differently and also this is not a clean OS installation and has been running for over 2 months, since getting the 3900x and is my daily driver. The Cinebench R15 multi threaded score is an average of 3002 over 4 tests. This score isn't great, but with real world workloads, the 200 points is pretty much meaningless if not then it's pretty damn close to being meaningless. The Cinebench R15 single threaded scored 199. I am also running my RAM overclocked to 3600 MHz with 16. 19. 19. 19. 38 timings. I can't do the timings any tighter, as they'll just crash the system and even freeze the BIOS (that was fun) This BIOS update has pretty much cured my itch of wanting to overclock. Sorry for the long comment, hopefully this also helps, someone although again, no where near as methodical as GN but the end results are similar with the Cinebench R15 scores and real world workloads. Build 3900x X570 Gigabyte Aorus Elite GTX 1080 Ti 2x 16 GB DDR4 3200 OC'd to 3600 On a custom loop. Thermal paste is MX 2. My Grizzly Kryonaut arrived a bit late.
reply
paranoidrodent
It works (for me at least! I got a 75 MHz bump to reach advertised max boost during testing. 8) 3900X owner with the same motherboard here. Just installed the new F7a BIOS. I am now getting two cores that peak at 4. 600GHz during testing (those capped at 4. 525 on earlier BIOSes) on my first chiplet (with the other 4 cores boosting to 4. 475-4. 525. So now every core on that chiplet is running as fast as my best ones were before, which is nice. All the cores of chiplet 1 seem to be getting better overall performance while chiplet 2 seems to be more or less the same. Single core tasks seem to just bounce around between the fast cores on chiplet 1. One of my cores bumped its highest value up to 4. 625GHz while idling after I did my stress testing (as I was typing this) so maybe I'm looking at a 100Mhz bump on that core (I haven't seen it while running a benchmark yet but it's nice to see the potential is there. My sustained single core boost is generally 4. 425-4. 500 with occasional peaks above that (up to 4. 6 but let's be honest, those are the peak spikes - 4. 475 is where it's averaging. My sustained all-12 core boost under load is 4. 1GHz across the board (3. 8 is base, same as before. Oddly enough, it was turning on PBO that got me to the 4. 6GHz mark. In default, I was capping at 4. 575 with the new BIOS (a 50MHz bump. PBO used to actually degrade my performance slightly before. Maybe they're starting to sort that thing out? Of course, given that 4. 625 blip, maybe I would have briefly hit 4. 6 on that one core without PBO if I had given it a few extra runs in Cinebench. I'm air-cooling using a NH-D15 SE-AM4 inside a Bequiet! case and tested with my case fans set for maximum air flow (touch noisier than my usual setting) so I doubt I'm getting quite as nominal a result as possible. With a good watercooling setup, I might squeeze a higher sustained boost out of it. Aside from turning on PBO on a lark afterwards, I tested using the methodology der8auer described in his video both before and after the BIOS update (Hardware Info and Cinebench R15 single core. No other changes to the system.
reply
It works (for me at least! I got a 75 MHz bump to reach advertised max boost during testing. 8) 3900X owner with the same motherboard here. Just installed the new F7a BIOS. I am now getting two cores that peak at 4. 600GHz during testing (those capped at 4. 525 on earlier BIOSes) on my first chiplet (with the other 4 cores boosting to 4. 475-4. 525. So now every core on that chiplet is running as fast as my best ones were before, which is nice. All the cores of chiplet 1 seem to be getting better overall performance while chiplet 2 seems to be more or less the same. Single core tasks seem to just bounce around between the fast cores on chiplet 1. One of my cores bumped its highest value up to 4. 625GHz while idling after I did my stress testing (as I was typing this) so maybe I'm looking at a 100Mhz bump on that core (I haven't seen it while running a benchmark yet but it's nice to see the potential is there. My sustained single core boost is generally 4. 425-4. 500 with occasional peaks above that (up to 4. 6 but let's be honest, those are the peak spikes - 4. 475 is where it's averaging. My sustained all-12 core boost under load is 4. 1GHz across the board (3. 8 is base, same as before. Oddly enough, it was turning on PBO that got me to the 4. 6GHz mark. In default, I was capping at 4. 575 with the new BIOS (a 50MHz bump. PBO used to actually degrade my performance slightly before. Maybe they're starting to sort that thing out? Of course, given that 4. 625 blip, maybe I would have briefly hit 4. 6 on that one core without PBO if I had given it a few extra runs in Cinebench. I'm air-cooling using a NH-D15 SE-AM4 inside a Bequiet! case and tested with my case fans set for maximum air flow (touch noisier than my usual setting) so I doubt I'm getting quite as nominal a result as possible. With a good watercooling setup, I might squeeze a higher sustained boost out of it. Aside from turning on PBO on a lark afterwards, I tested using the methodology der8auer described in his video both before and after the BIOS update (Hardware Info and Cinebench R15 single core. No other changes to the system.
reply
Krazi
its not AMD's fault if each model and sure as hell not each individual CPU ( say all 3700X or 3900X chips in the world( 3600/3600X, 3700X, 3800X, 3900X and soon the 3950X 16C/) cant hit the 1 core/thread advertised boost clocks. its the silicon the chip is birthed from, not every piece of silicon is gonna be the same either! its almost impossible to make millions of CPU dies, let alone millions or billions of the different model CPU's that there are. their not all gonna be 1: 1 replication. they can be Very Close by the Binning Process, but that can only be so close 1: 2 ratio maybe. or you get limited amount of say the 3900X which is small stock as is. i cant even find one in stock any where and the 3950X will sell out the first day its out and there will probably be a flood of old 3900X's cuz every one sold their 3900X 12 core to buy the 3950X 16 core
reply
its not AMD's fault if each model and sure as hell not each individual CPU ( say all 3700X or 3900X chips in the world( 3600/3600X, 3700X, 3800X, 3900X and soon the 3950X 16C/) cant hit the 1 core/thread advertised boost clocks. its the silicon the chip is birthed from, not every piece of silicon is gonna be the same either! its almost impossible to make millions of CPU dies, let alone millions or billions of the different model CPU's that there are. their not all gonna be 1: 1 replication. they can be Very Close by the Binning Process, but that can only be so close 1: 2 ratio maybe. or you get limited amount of say the 3900X which is small stock as is. i cant even find one in stock any where and the 3950X will sell out the first day its out and there will probably be a flood of old 3900X's cuz every one sold their 3900X 12 core to buy the 3950X 16 core
reply
stevin47
could you help me. asus tuf gaming x570 plus (wifi) current ver bios 1201 no update containing bios agesia 1. 0. 0. 4. update. the asus tuf gaming x570 plus (none wifi) has had the update over a wk to bios ver 1202 have had so much problems the tuf wifi mb and cpu clocks can not even set to performance mode in bio's without freezes and crashes. asus response is we will get back to you been days nothing / also got the board in aug and never been able to register the asus motherboard input serial # and tells me incorrect # so used command prompt to get serial number same thing incorrect # in bios mouse so slow and hard to move around was hoping the update would fix these things the boards are the same but one has wifi and diff bios why asus is not updating it or even giving me help on this matter?
reply
could you help me. asus tuf gaming x570 plus (wifi) current ver bios 1201 no update containing bios agesia 1. 0. 0. 4. update. the asus tuf gaming x570 plus (none wifi) has had the update over a wk to bios ver 1202 have had so much problems the tuf wifi mb and cpu clocks can not even set to performance mode in bio's without freezes and crashes. asus response is we will get back to you been days nothing / also got the board in aug and never been able to register the asus motherboard input serial # and tells me incorrect # so used command prompt to get serial number same thing incorrect # in bios mouse so slow and hard to move around was hoping the update would fix these things the boards are the same but one has wifi and diff bios why asus is not updating it or even giving me help on this matter?
reply
Tasos
Hey Steve I do not really expect you to look into older threads/videos so I am not expecting a reply but I will post this here because this agesa update gave me ram issues. I am still on a b350 board/ Gigabyte AB Gaming 3. Gigabyte have been good with bios updates although at a later time compared with the x570 boards. The latest F42d with the 1. 0. 0. 3 ABBA code unfortunately gives me dram related BSOD's like fltmgr. sys or page fault in nonpaged area errors. I am running 4x8GB Corsair 3000 CL16 2 with micron and 2 with samsung chips. My 3600 with PBO on stock voltages. Reverted back to F42a with older 1. 0. 0. 3 code and problem disappeared. Anything similar anyone? BTW all sticks are fine/ checked with memtest
reply
Hey Steve I do not really expect you to look into older threads/videos so I am not expecting a reply but I will post this here because this agesa update gave me ram issues. I am still on a b350 board/ Gigabyte AB Gaming 3. Gigabyte have been good with bios updates although at a later time compared with the x570 boards. The latest F42d with the 1. 0. 0. 3 ABBA code unfortunately gives me dram related BSOD's like fltmgr. sys or page fault in nonpaged area errors. I am running 4x8GB Corsair 3000 CL16 2 with micron and 2 with samsung chips. My 3600 with PBO on stock voltages. Reverted back to F42a with older 1. 0. 0. 3 code and problem disappeared. Anything similar anyone? BTW all sticks are fine/ checked with memtest
reply
IncertusetNescio
Yeah I'm not entirely sure what's happening with Ryzen Gen 3 and its clock management and why. My 3700x idles at 4. 2 GHz and like 60-65C and almost never goes lower than that. on a fairly recent (for MSI, non-Beta) BIOS. Not sure if I want to update it until the clock fix one comes out after the current Beta one. Meanwhile I'm clock-checking it with Ryzen Master to stock at a nice low voltage just to keep it from revving all the time. For reference, the previous chip, the 1700, idled normally for its speed range and at 45-50C, and at a sane voltage. Same case, same cooler, identical fan profile. The 3700x IS a 65W chip. right?
reply
Yeah I'm not entirely sure what's happening with Ryzen Gen 3 and its clock management and why. My 3700x idles at 4. 2 GHz and like 60-65C and almost never goes lower than that. on a fairly recent (for MSI, non-Beta) BIOS. Not sure if I want to update it until the clock fix one comes out after the current Beta one. Meanwhile I'm clock-checking it with Ryzen Master to stock at a nice low voltage just to keep it from revving all the time. For reference, the previous chip, the 1700, idled normally for its speed range and at 45-50C, and at a sane voltage. Same case, same cooler, identical fan profile. The 3700x IS a 65W chip. right?
reply
Googlar
It is not just possible, but actually LIKELY, that AMD knew about this issue before anybody else did. If that's true, then presumably they could have delayed the launch in order to fix it, but I think everyone can agree that delaying the launch just to make sure that every single cpu was able to hit advertised max boost clocks would have been really unnecessary, given that we're talking about a very small impact upon performance. In any case, it seems that the issue is fixed now, though I guess we'll need to get confirmation from a bunch of different sources on a bunch of different CPUs before we can conclude that with confidence.
reply
It is not just possible, but actually LIKELY, that AMD knew about this issue before anybody else did. If that's true, then presumably they could have delayed the launch in order to fix it, but I think everyone can agree that delaying the launch just to make sure that every single cpu was able to hit advertised max boost clocks would have been really unnecessary, given that we're talking about a very small impact upon performance. In any case, it seems that the issue is fixed now, though I guess we'll need to get confirmation from a bunch of different sources on a bunch of different CPUs before we can conclude that with confidence.
reply
Tech
Just Flashed back to this Bios AGESA 1003 ABBA Code (took a day to find the old now pulled bios. MSI x370 pro carbon and probably most 300 series msi boards have a problem with AGESA 1004 B beta bios. Reports 10c hotter than it really is, so doesn't boost correctly when testing in fact it actively protects itself to aggressively. This is the big problem of what I perceived my performance should be according to AMD promotions and what I actually got on a beta bios. I wish MSI would test better before releasing a bios to the general public. Worst thing is MSI communication is BAD.
reply
Just Flashed back to this Bios AGESA 1003 ABBA Code (took a day to find the old now pulled bios. MSI x370 pro carbon and probably most 300 series msi boards have a problem with AGESA 1004 B beta bios. Reports 10c hotter than it really is, so doesn't boost correctly when testing in fact it actively protects itself to aggressively. This is the big problem of what I perceived my performance should be according to AMD promotions and what I actually got on a beta bios. I wish MSI would test better before releasing a bios to the general public. Worst thing is MSI communication is BAD.
reply
Red-Dragon11
I think people should be concerned when a company claims something publically only to find out that in real world usage this is far from the case. This is not a criticism only at AMD but you think from past performances AMD would have learned. It would be great if more companies decided to take a stance and not follow everyone else and actually be totally transparent, imo this give better credibility in the long run. Great update this has been a long time coming well in fact this review update should never have had to happen at all.
reply
I think people should be concerned when a company claims something publically only to find out that in real world usage this is far from the case. This is not a criticism only at AMD but you think from past performances AMD would have learned. It would be great if more companies decided to take a stance and not follow everyone else and actually be totally transparent, imo this give better credibility in the long run. Great update this has been a long time coming well in fact this review update should never have had to happen at all.
reply
WarDevil95
3900x Asus Crosshair VIII x570 Wi Fi and I just flashed in the 1001 PI 1. 0. 0. 3 patch to approve Abba last night and it has made no difference in clock speeds. Running Cinebench 20 with Ryzan master on core 1 I get 43xx MHz and core two 4200mhz and every core after that stays at 4. 0ghz. Tried enabling pbo instead of auto and it changed nothing. Did I get a dud of a chip? Would like to know by Friday as I live 2 hours away from the city and I am going up there I could swing by and get a replacement at memory express.
reply
3900x Asus Crosshair VIII x570 Wi Fi and I just flashed in the 1001 PI 1. 0. 0. 3 patch to approve Abba last night and it has made no difference in clock speeds. Running Cinebench 20 with Ryzan master on core 1 I get 43xx MHz and core two 4200mhz and every core after that stays at 4. 0ghz. Tried enabling pbo instead of auto and it changed nothing. Did I get a dud of a chip? Would like to know by Friday as I live 2 hours away from the city and I am going up there I could swing by and get a replacement at memory express.
reply
Add a review, comment
Other channel videos
