Trying to understand Ryzen 3000 series boost speed variations

[tabascosauz]

I remember a video by BZ where he went through all the BIOS settings in the gigabyte X470 board and I think he said the max offset the board allowed was 300, I think that was mv but I'm not sure. I have it bookmarked downstairs though.

At some point I'd like to run through an offset with you so that I understand what is required, if that's ok with you?

It's your thread - you don't have to ask those of us that are here are here because it's fun to hang out here lol

Perhaps -0.3V is what he meant by the maximum offset allowed? In any case, that's too much of an offset, either way - positive or negative. If you set that much of an offset, performance will drop for sure, and it may be unstable. You shouldn't have to go past -0.1V. As a general rule, start from 0, work your way down in the increments it allows (I think it's like 0.0075V steps or something that small), and make sure that it's actually changing your Vcore (I think some of the values are somewhat redundant, they result in the same Vcore as the setting just above or below them, like -0.06875 and -0.075 are the same). Find a spot that's a good balance between performance/heat/noise. For me, -0.05V is good, but -0.075V is actually still where I bench highest on CB. A lot of boards will feed too much Vcore stock on Auto voltage, which seems to be detrimental as it generates too much heat, and hurts performance by hitting the limiter all the time.

I think I figured out -0.075V was stable through a couple of hours of P95 Smallest back in September or so, and just kept at it all this time. I think -0.08V started logging WHEA errors in HWInfo (WHEA errors kinda like a friendly "hey, I think this will run into problems down the road" reminder for me, followed up in order of severity by P95 crashing, OCCT crashing, BSODs, graphical artifacting, and finally failed boot), so I stayed at -0.075V.

The Core Voltage field and offset setting seems to be really janky; Auto setting is default and doesn't allow you to set offset as it's greyed out. If you change Voltage to Normal, you can then set an offset. You can change it back to Auto if ever you wish to go back to stock Vcore, however, the offset doesn't go away, it somehow stays even if you're back on Auto (and you can't change or remove the offset on Auto, it's greyed out even though it's there). Like, if you run the system on Auto again but didn't manually reset the offset to 0, boot into Windows with it on Auto voltage and it'll still act like it's offset. So if you ever want to change the Vcore offset back, you have to be in Normal mode first, remove it, then switch to Auto mode, and you'll be back on stock Vcore.

 

[Waldorf]

I read someone mentioned CB15 (?) in a 300s loop was one of the quickest ways to get to max boost clocks (pass=stable).
not sure if he himself or others well and just posting) but i know they did it on a couple of different chips to verify.
at least faster than running prime (which iirc is not always a 100% guarantee for stable clocks outside prime).

anyone knows what a 3600 can usually do on all cores?
trying to push clocks a bit, but not to squeeze everything out of it.

 

[lorry]

It's your thread - you don't have to ask those of us that are here are here because it's fun to hang out here lol

Perhaps -0.3V is what he meant by the maximum offset allowed? In any case, that's too much of an offset, either way - positive or negative. If you set that much of an offset, performance will drop for sure, and it may be unstable. You shouldn't have to go past -0.1V. As a general rule, start from 0, work your way down in the increments it allows (I think it's like 0.0075V steps or something that small), and make sure that it's actually changing your Vcore (I think some of the values are somewhat redundant, they result in the same Vcore as the setting just above or below them, like -0.06875 and -0.075 are the same). Find a spot that's a good balance between performance/heat/noise. For me, -0.05V is good, but -0.075V is actually still where I bench highest on CB. A lot of boards will feed too much Vcore stock on Auto voltage, which seems to be detrimental as it generates too much heat, and hurts performance by hitting the limiter all the time.

I think I figured out -0.075V was stable through a couple of hours of P95 Smallest back in September or so, and just kept at it all this time. I think -0.08V started logging WHEA errors in HWInfo (WHEA errors kinda like a friendly "hey, I think this will run into problems down the road" reminder for me, followed up in order of severity by P95 crashing, OCCT crashing, BSODs, graphical artifacting, and finally failed boot), so I stayed at -0.075V.

The Core Voltage field and offset setting seems to be really janky; Auto setting is default and doesn't allow you to set offset as it's greyed out. If you change Voltage to Normal, you can then set an offset. You can change it back to Auto if ever you wish to go back to stock Vcore, however, the offset doesn't go away, it somehow stays even if you're back on Auto (and you can't change or remove the offset on Auto, it's greyed out even though it's there). Like, if you run the system on Auto again but didn't manually reset the offset to 0, boot into Windows with it on Auto voltage and it'll still act like it's offset. So if you ever want to change the Vcore offset back, you have to be in Normal mode first, remove it, then switch to Auto mode, and you'll be back on stock Vcore.

I ask because that's manners, you owe me nothing

Yeah, that's what I meant, plus or minus 0.3v. He was just saying what the range was. The BIOS run-through is actually on a Gigabyte X470 Ultra Gaming , but the BIOS is virtually identical and good enough as an example.
Which would be the best to use to check ? HWInfo or RM?
Is it the CPU Core Voltage (SVI2 TFN) in HWInfo? That, at desktop sits around 1.1-1.4V, Lowest is 1.094, highest is 1.481
"To change the Vcore offset back, you have to be in Normal mode first, remove it, then switch to Auto mode, and you'll be back on stock Vcore." - good to know, I may well have missed that

I read someone mentioned CB15 (?) in a 300s loop was one of the quickest ways to get to max boost clocks (pass=stable).
not sure if he himself or others well and just posting) but i know they did it on a couple of different chips to verify.
at least faster than running prime (which iirc is not always a 100% guarantee for stable clocks outside prime).

anyone knows what a 3600 can usually do on all cores?
trying to push clocks a bit, but not to squeeze everything out of it.

They say 4.3 here at 1.43V

AMD Ryzen 5 3600 CPU Review & Benchmarks: Strong Recommendation from GN | GamersNexus

stub Alongside the 3900X and 3700X that we’re also reviewing, AMD launched its R5 3600 today to the public. We got a production sample of one of the R5 3600 CPUs through a third-party and, after seeing its performance, we wanted to focus first on this one for our initial Ryzen 3000 review. We’ve...

www.gamersnexus.net

 

[tabascosauz]

I read someone mentioned CB15 (?) in a 300s loop was one of the quickest ways to get to max boost clocks (pass=stable).
not sure if he himself or others well and just posting) but i know they did it on a couple of different chips to verify.
at least faster than running prime (which iirc is not always a 100% guarantee for stable clocks outside prime).

anyone knows what a 3600 can usually do on all cores?
trying to push clocks a bit, but not to squeeze everything out of it.

I could have sworn I've seen someone hitting 4.3 or 4.4 over on the Zen Owner's Garden thread. Fixed freq of course. It's easy enough to test your chip's limits with manual clocks and Vcore. Minimum silicon quality on 3600 sucks butt.

Smallest FFTs is alright, but Small appears no longer viable because it seems to artificially reduce the multiplier. The nice thing about something like Smallest is that it'll be pretty much the max load possible, meaning Vcore droops to its maximum. Even at maximum droop, there is some fluctuation, and it is those tiny lows that quickly kill worker threads and therefore stability if you're right on the edge.

I ask because that's manners, you owe me nothing

Yeah, that's what I meant, plus or minus 0.3v. He was just saying what the range was. The BIOS run-through is actually on a Gigabyte X470 Ultra Gaming , but the BIOS is virtually identical and good enough as an example.
Which would be the best to use to check ? HWInfo or RM?
Is it the CPU Core Voltage (SVI2 TFN) in HWInfo? That, at desktop sits around 1.1-1.4V, Lowest is 1.094, highest is 1.481
"To change the Vcore offset back, you have to be in Normal mode first, remove it, then switch to Auto mode, and you'll be back on stock Vcore." - good to know, I may well have missed that

Yes, SVI2 TFN is reported by the CPU and is the most accurate. The other Vcore you'll see is reported by your board's ITE chip, and is not as useful. RM has been really glitchy in the past (there was this period of time when it would report Vcore as a constant 1.4V+ as long as you had the histogram display enabled in the application), so just use HWInfo.

I think I see the source of possible confusion here. Visually, all 400-series Gigabytes have the same BIOS (and even X570, just kinda reskinned and reshuffled here and there), but the pre-Matisse BIOS versions look a bit different to the post-Matisse release versions. I built a 2600 (Pinnacle) system for a friend with the exact same board as mine, but with an earlier BIOS (F4 or F5, I think), and he has things like EZ Overclock Tuner in BZ's video, which they seem to have taken away for Matisse, as it's not there in F42c, F42g, F50a and F50.



I'm guessing your F50 should like something like this? BZ's Pinnacle BIOS doesn't have a separate CPU Vcore field, which is helpful if you want to run fixed freq but also introduces the whole Normal/Auto headache. There used to be a functional difference between the two, not anymore it seems.

Whether you have anything to gain kinda depends on the chip you have, but that 1.481 Vcore peak is in line with stock highs on mine, though mine goes all the way up to 1.5V without any offset.

 

[Zach_01]

Yesterday found what FIT stands for... thanks to the link @lorry provided.
All and all it’s about FITness silicon manager. What we call the internal regulator/manager of the CPU, the one that controls and regulate the PPT/TDC/EDC and temperature.

I also tried to undervolt my 3600 with offset (vcore normal and offset a couple of notches, -0.00625 each =0.0125 under normal)
I did see a regression in max temp, about 2C but also loosing some clock. At least 25MHz.
I guess I hit the lottery of the worst possible silicon quality and cannot sustain clocks even with the slightest undervolting.

My 3600 at auto vcore clocks to about 4200MHz with short bursts for single/low core load, and about 4000MHz for all core load.

I assume that higher silicon quality CPUs would benefit from a slight undervolt. At least to preserve same clocks with lower powerdraw and temp

 

[lorry]

Smallest FFTs is alright, but Small appears no longer viable because it seems to artificially reduce the multiplier. The nice thing about something like Smallest is that it'll be pretty much the max load possible, meaning Vcore droops to its maximum. Even at maximum droop, there is some fluctuation, and it is those tiny lows that quickly kill worker threads and therefore stability if you're right on the edge.



Yes, SVI2 TFN is reported by the CPU and is the most accurate. The other Vcore you'll see is reported by your board's ITE chip, and is not as useful. RM has been really glitchy in the past (there was this period of time when it would report Vcore as a constant 1.4V+ as long as you had the histogram display enabled in the application), so just use HWInfo.

I think I see the source of possible confusion here. Visually, all 400-series Gigabytes have the same BIOS (and even X570, just kinda reskinned and reshuffled here and there), but the pre-Matisse BIOS versions look a bit different to the post-Matisse release versions. I built a 2600 (Pinnacle) system for a friend with the exact same board as mine, but with an earlier BIOS (F4 or F5, I think), and he has things like EZ Overclock Tuner in BZ's video, which they seem to have taken away for Matisse, as it's not there in F42c, F42g, F50a and F50.

View attachment 140915

I'm guessing your F50 should like something like this? BZ's Pinnacle BIOS doesn't have a separate CPU Vcore field, which is helpful if you want to run fixed freq but also introduces the whole Normal/Auto headache. There used to be a functional difference between the two, not anymore it seems.

Whether you have anything to gain kinda depends on the chip you have, but that 1.481 Vcore peak is in line with stock highs on mine, though mine goes all the way up to 1.5V without any offset.

I'm sure that I still have the 'EZ Overclock Tuner' option in my BIOS, I'll try to remember next time i'm in there and screenshot, but your pic does look familiar
Do still kind of think though that I have little to actually gain re CPU speeds overall. I mean yes I Might gain slightly in one thing, then lose that in another, so that overall most things remain the same.

Yesterday found what FIT stands for... thanks to the link @lorry provided.
All and all it’s about FITness silicon manager. What we call the internal regulator/manager of the CPU, the one that controls and regulate the PPT/TDC/EDC and temperature.

I also tried to undervolt my 3600 with offset (vcore normal and offset a couple of notches, -0.00625 each =0.0125 under normal)
I did see a regression in max temp, about 2C but also loosing some clock. At least 25MHz.
I guess I hit the lottery of the worst possible silicon quality and cannot sustain clocks even with the slightest undervolting.

My 3600 at auto vcore clocks to about 4200MHz with short bursts for single/low core load, and about 4000MHz for all core load.

I assume that higher silicon quality CPUs would benefit from a slight undervolt. At least to preserve same clocks with lower powerdraw and temp

Cool, FIT makes sense Now, lol.

I'll have to see if mine benefits, but not today

Today is about consuming silly amounts of 'fuel'

 

[Waldorf]

Seems i do get what i paid for (3.95-4 ghz on all core load/4.2 for low load), so i dont expect it to go much higher.
Should have kept the last one that i swapped to try this one, as that got 50-200 MHz more

 

[lorry]

Seems i do get what i paid for (3.95-4 ghz on all core load/4.2 for low load), so i dont expect it to go much higher.
Should have kept the last one that i swapped to try this one, as that got 50-200 MHz more

Ouch

 

[tabascosauz]

Yesterday found what FIT stands for... thanks to the link @lorry provided.
All and all it’s about FITness silicon manager. What we call the internal regulator/manager of the CPU, the one that controls and regulate the PPT/TDC/EDC and temperature.

I also tried to undervolt my 3600 with offset (vcore normal and offset a couple of notches, -0.00625 each =0.0125 under normal)
I did see a regression in max temp, about 2C but also loosing some clock. At least 25MHz.
I guess I hit the lottery of the worst possible silicon quality and cannot sustain clocks even with the slightest undervolting.

Pretty much, your chip has to be pulling excessive stock voltage, and also getting too hot on stock voltage for the cooler you're using, to get anything out of undervolting. You've got plenty of cooling, and water isn't afraid of a little extra voltage, so go for performance!

It's mostly a SFF thing. The U9S and D9L are on the small side, but there are a lot of other miserable little coolers in these small cases that make those two 92mm towers look like kings.

If you'd like, you can test with fixed freq. It gives you a much clearer picture as to how you did in the lottery. As far as I can tell, the 95c throttle limit is the only safeguard with full manual settings.

I'm sure that I still have the 'EZ Overclock Tuner' option in my BIOS, I'll try to remember next time i'm in there and screenshot, but your pic does look familiar
Do still kind of think though that I have little to actually gain re CPU speeds overall. I mean yes I Might gain slightly in one thing, then lose that in another, so that overall most things remain the same.

Cool, FIT makes sense Now, lol.

I'll have to see if mine benefits, but not today

Today is about consuming silly amounts of 'fuel'

A lot of those 1-click OC utilities use obscene amounts of voltage to clock higher (not that it's really possible on Matisse anyway). I'm glad it appears to be removed.

Happy new year, all.

 

[lorry]

Pretty much, your chip has to be pulling excessive stock voltage, and also getting too hot on stock voltage for the cooler you're using, to get anything out of undervolting. You've got plenty of cooling, and water isn't afraid of a little extra voltage, so go for performance!

It's mostly a SFF thing. The U9S and D9L are on the small side, but there are a lot of other miserable little coolers in these small cases that make those two 92mm towers look like kings.

If you'd like, you can test with fixed freq. It gives you a much clearer picture as to how you did in the lottery. As far as I can tell, the 95c throttle limit is the only safeguard with full manual settings.



A lot of those 1-click OC utilities use obscene amounts of voltage to clock higher (not that it's really possible on Matisse anyway). I'm glad it appears to be removed.

Happy new year, all.

Erm, you was saying?



I played, mind you a lil tip. do not mess with ya BIOS when inebriated, took me Way too long to get the bladdy thing to boot properly again (by properly I mean having the MB RGB working, lol);






Nappy new year all

Just out of interest, has any of you tried this and is it any good?

https://www.amd.com/en/technologies/store-mi

Oh and watching a BZ video sometimes makes me wish that my brain was watercooled!

 

[NoJuan999]

The idle temps in RM were consistently around 5c higher when using the 1usmus plan on 1903.
I just manually updated to 1909 so I will see if that is still the case now and post back.

Update:
With Win 10 1909 and the 1usmus Universal Plan active (with minimum processor state at 20% and max at 100%) my temps are now back in line with what I had with 1903/AMD Ryzen Balanced Plan (same Processor state settings).
I will say that it hasn't made any difference in my CB R20 scores though (multi or single).
And the same cores I saw being utilized with the Ryzen Balanced Pan (and 1903) are still the cores being used by 1909 and the 1usmus plan.

If I had to guess, I'd say that is because my 2 best cores (according to HWinfo) seem to be core 0 (core 1 with silver circle in RM) and core 3 (core 4 with silver star in RM) which are both on the first CCX.
RM shows the first core (RM core 5, HWinfo core 4) on CCX 2 as the best overall core though (gold star).

And HWinfo has shown that all 3 of those cores hit 4.4 GHZ.

 

[tabascosauz]

Just out of interest, has any of you tried this and is it any good?

https://www.amd.com/en/technologies/store-mi

Oh and watching a BZ video sometimes makes me wish that my brain was watercooled!

StoreMI is basically AMD's version of Intel's Smart Response Technology (SRT) that's been around for a while. Basically takes your SSD and HDD and combines them to make a DIY SSHD of sorts, and visible as a single drive. From the wording, it seems StoreMI can also cache SATA SSDs with NVMe. Neither have ever been that popular, especially that SSDs have gotten so cheap.

 

[lorry]

StoreMI is basically AMD's version of Intel's Smart Response Technology (SRT) that's been around for a while. Basically takes your SSD and HDD and combines them to make a DIY SSHD of sorts, and visible as a single drive. From the wording, it seems StoreMI can also cache SATA SSDs with NVMe. Neither have ever been that popular, especially that SSDs have gotten so cheap.

Yeah I knew that had been around for a while, just wondered if they worked properly. About the only advantage for myself might be the ability to add in some ram, seeing as I have 32GB.
The two drives in here are both SSD (Samsung Evo 970 1TB NVMe M.2 SSD & Intel 660P 2TB NVMe M.2 SSD).
I Was going to get a large HDD but have now decided that my external 3TB USB drive is enough for now as a backup, as it came be easily disconnected after the weekly backup, so totally removed fron any potential virus / ransonware threats.

 

[tabascosauz]

Yeah I knew that had been around for a while, just wondered if they worked properly. About the only advantage for myself might be the ability to add in some ram, seeing as I have 32GB.
The two drives in here are both SSD (Samsung Evo 970 1TB NVMe M.2 SSD & Intel 660P 2TB NVMe M.2 SSD).
I Was going to get a large HDD but have now decided that my external 3TB USB drive is enough for now as a backup, as it came be easily disconnected after the weekly backup, so totally removed fron any potential virus / ransonware threats.

Welcome to the all-SSD club I have a Seagate 2TB external that kinda just accumulates junk that I don't want on the 2.75TB of SSDs. It's worked pretty well over the years

Aside from the little-documented caching of SATA SSD with NVMe SSD, this kind of caching has really never worked that well. I had a number of experiences with Intel SRT, working as it was designed with a 500GB-1TB HDD and a small 32GB mSATA SSD. It was marginally faster at first, but after a few months it was pretty much indistinguishable from HDD performance, and they all ended up being upgraded as all-SSD setups - much faster.

Another caveat is that this kind of caching is designed for and most prevalent in laptops. I suspect this is because laptops aren't susceptible to power loss from blackouts and the like, unlike non-UPS protected desktops. Same goes for using RAM caching for SSDs (which Samsung has been fond of for years as a Samsung Magician feature). Can you imagine what ugly consequences await when a SRT or RAM cached system suddenly loses power with data currently being moved in and out of cache? *shudder*

There were some people using small (insanely fast) Optane drives to accelerate slower SSDs, but again, very limited market. Since the 660p is a relatively (to 3D TLC) slow drive outside of its limited cache, I guess there might be a small use for caching again. But I'd tread carefully around QLC, so maybe not.

 

[lorry]

Welcome to the all-SSD club I have a Seagate 2TB external that kinda just accumulates junk that I don't want on the 2.75TB of SSDs. It's worked pretty well over the years

Aside from the little-documented caching of SATA SSD with NVMe SSD, this kind of caching has really never worked that well. I had a number of experiences with Intel SRT, working as it was designed with a 500GB-1TB HDD and a small 32GB mSATA SSD. It was marginally faster at first, but after a few months it was pretty much indistinguishable from HDD performance, and they all ended up being upgraded as all-SSD setups - much faster.

Another caveat is that this kind of caching is designed for and most prevalent in laptops. I suspect this is because laptops aren't susceptible to power loss from blackouts and the like, unlike non-UPS protected desktops. Same goes for using RAM caching for SSDs (which Samsung has been fond of for years as a Samsung Magician feature). Can you imagine what ugly consequences await when a SRT or RAM cached system suddenly loses power with data currently being moved in and out of cache? *shudder*

There were some people using small (insanely fast) Optane drives to accelerate slower SSDs, but again, very limited market. Since the 660p is a relatively (to 3D TLC) slow drive outside of its limited cache, I guess there might be a small use for caching again. But I'd tread carefully around QLC, so maybe not.

Yeah I wasn't seriously considering it, more a case of just wondering about it. Everything loads plenty fast enough for me so I'm not desperate for however few seconds it may or may not shave off.

An UPS Is something I want get, although we do only get a couple of power cuts a year.

 

[Waldorf]

helps with psu life (no voltage fluctuations) as well, but biggest gain is brown-outs or when going out of range for the psu.
can say about once a month i can hear the UPS "click" when it kicks in to regulate the voltage, usually doesnt even need to switch to battery (just using AVR).
really nice as a way to prevent machine from crashing when ur not home (shutdown feat), and works as a backup to charge phone etc if power is out for a while

 

[Zach_01]

Happy New year!!!

Here is about AMD’s storeMI. Works pretty well for those who can’t afford many or large SSD...

 

[tabascosauz]

Yeah I wasn't seriously considering it, more a case of just wondering about it. Everything loads plenty fast enough for me so I'm not desperate for however few seconds it may or may not shave off.

An UPS Is something I want get, although we do only get a couple of power cuts a year.

That was my reasoning for holding off on a UPS for so long. It used to be a once or twice a year occurrence, and was fixed relatively quickly; within 3 hours at the most. Worst case, I'd just sit in Starbucks up the street with my XPS until power came back on.

This year, things were particularly bad. Blackouts became a monthly or bimonthly occurrence, with one stint where some transformer at a major substation a ways away kept exploding (videos taken by highway commuting people showed some pretty spectacular fireworks going up) three days in a row. Three daily blackouts in a row for tens of thousands of households. The second one wasn't fixed until more than 8 hours later. Had a earsplitting lightning strike pretty much right down the block nuke Internet for most of the day, too.

I had some weird corruption of Windows that kinda lurked in the background and finally gave up the ghost while I was testing timings on the new 32GB DJR kit. Straight up just gave up. Had to reinstall. Sucky thing was that actually occurred after I bought the BR1500MS, but the damage was probably done after months of these power grid events. It's only been a little while with the new APC unit, but it's already saved me from a brownout and a quick blackout (it's become concerningly common to have these 0.5-1 second blackouts from which the PC seems to recover without restarting).

helps with psu life (no voltage fluctuations) as well, but biggest gain is brown-outs or when going out of range for the psu.
can say about once a month i can hear the UPS "click" when it kicks in to regulate the voltage, usually doesnt even need to switch to battery (just using AVR).
really nice as a way to prevent machine from crashing when ur not home (shutdown feat), and works as a backup to charge phone etc if power is out for a while

Spot on. Though, now that I have a UPS I actually got more into the habit of putting my PC to sleep when I leave. It wakes quickly, but also only draws 1-3W so if the power does go out, I don't have to lose potential data through a forced shutdown after a certain period of time, nor do I deplete the UPS.

 

[lorry]

helps with psu life (no voltage fluctuations) as well, but biggest gain is brown-outs or when going out of range for the psu.
can say about once a month i can hear the UPS "click" when it kicks in to regulate the voltage, usually doesnt even need to switch to battery (just using AVR).
really nice as a way to prevent machine from crashing when ur not home (shutdown feat), and works as a backup to charge phone etc if power is out for a while

I have a 1000W Titanium PSU, hopefully that will never top out, at least not anytime soon anyway! I decided on that size in the hope that it would hopefully be good for a couple of generations at least.
When I first moved here we got up to a dozen power cuts a year, as our sub-station wasn't really big enough to handle all the homes power needs once the new roads and homes had been built (I live on the edge of a 'new' town, one built After WWII). They finally upgraded it and since then the power cuts have dropped to a couple a year at the max

That was my reasoning for holding off on a UPS for so long. It used to be a once or twice a year occurrence, and was fixed relatively quickly; within 3 hours at the most. Worst case, I'd just sit in Starbucks up the street with my XPS until power came back on.

Spot on. Though, now that I have a UPS I actually got more into the habit of putting my PC to sleep when I leave. It wakes quickly, but also only draws 1-3W so if the power does go out, I don't have to lose potential data through a forced shutdown after a certain period of time, nor do I deplete the UPS.

As I said above, my situation is the reverse, we have gone from at least one a month to twice a year, at the very most.

As I know nothing about UPS apart from the little that I have read, I would like to pick your collective knowledge if I may.
Outervision calculate my power needs as follows (monitor included)

Power Supply Calculator - PSU Calculator | OuterVision

Power Supply Calculator - Select computer parts and our online PSU calculator will calculate the required power supply wattage and amperage for your PC.

outervision.com

Load Wattage: 619 W
Recommended UPS rating: 1250 VA
Recommended PSU Wattage: 669 W

However, I have never seen my Entire home's wattage usage go above 500, even when stress testing.

I know nothing about brands but have looked at these two as possibilities IF they are any good of course!

Powercool 1200VA Uninterruptable Power Supply

Buy from Scan - 1200VA Powercool UPS, 3x UK Plug, 3x Female Kettle Lead Ports, 2x RJ45, USB Type-B, Built in AVR, Audible Alarm

www.scan.co.uk

APC 1400Va 700W UPS Line Interactive 6x IEC Sockets

Buy from Scan - 1400Va APC BX1400UI Mini Tower UPS 700W Line Interactive 6x IEC Sockets Black

www.scan.co.uk

IF they aren't a good type/brand can you guys make some suggestions as to what would be better?

One thing I just Don't get though is that R20 does Not keep my all time best scores. It Only keeps the best score for That session that my pc is switched on and yes 'keep best score' is ticked ??

Happy New year!!!

Here is about AMD’s storeMI. Works pretty well for those who can’t afford many or large SSD...

As you say it seems to be useful for those that can't afford large SSD.
What might have persuaded me to try it out would have been being able to use more than 2GB of ram. Having 32GB and rarely seeing more than 10GB in use, I could have easily slapped say 10GB of ram into a 'new drive' Now That technology would have been something to shout about.
But having said all that, just How much of a hardship is to have to wait a minute or so to load a game? I guess that in this 'instant world' that we live in today, some would call it a necessity.

 

[Deleted member 185158]

But having said all that, just How much of a hardship is to have to wait a minute or so to load a game? I guess that in this 'instant world' that we live in today, some would call it a necessity.

The young guys have far less patients than older gentlemen.

Before SSDs, Running Raid 0 on 10,000 rpm raptors was the way to go. Load times matter.

 

[lorry]

The young guys have far less patients than older gentlemen.

Before SSDs, Running Raid 0 on 10,000 rpm raptors was the way to go. Load times matter.

Kids These days, they need to learn some patience !

 

[Zach_01]

But it’s not only about game load times. Everything go/start faster from app start, OS boot, win explorer opening.
The system feels snappier all together when using SSD. Real or storeMI...

Back then we all used to slow times because that was it. We didn’t have smartphones or tablets and we did know how a fast system can make a difference. Now even me, can’t go back and I want SSD like performance. I want my phone/tablet to be reasonably fast, and my PC too...

 

[lorry]

But it’s not only about game load times. Everything go/start faster from app start, OS boot, win explorer opening.
The system feels snappier all together when using SSD. Real or storeMI...

Back then we all used to slow times because that was it. We didn’t have smartphones or tablets and we did know how a fast system can make a difference. Now even me, can’t go back and I want SSD like performance. I want my phone/tablet to be reasonably fast, and my PC too...

oh agreed, i also remember those times, my first build was just after we moved here in '93

so UPS ?

@lorry I got you beat on that ST score, don't ask me how, I've no idea :

View attachment 140583

I guess 1usmus must be putting my Cinebench ST test on Core 5 and 7, because if it was on Core 0 then I'd be way down in Zen+ and Intel territory. But 7300pts is very respectable for your 3900X. 500pts is also pretty much at the top of the pack, so could be down to chip variance or normal benchmark variation.

I feel that on reflection, one bios cannot be completely compatible with Every single motherboard out there.

You are undoubtedly correct; however, Gigabyte evidently has a different opinion on that matter. Though in all fairness to them, all the vendors do it to some extent - no sense in writing something new from the ground up for every board of a given generation. Then again, it's Gigabyte, so they're copy-pasting hot garbage in place of a polished and verified BIOS.

Haven't heard anything regarding the next iteration of AGESA for Matisse, so you may be stuck with F50 for a while. Is it possible to downgrade Gigabyte BIOSes?


@NoJuan999 you are not on the "latest" Windows, 1909 has a new scheduler that v1.1 of 1usmus is designed to work with; you're supposed to use the Universal plan if you don't have 3000 on 1909, and Ryzen Power Plan if you have 3000 on 1909. Also, "2.2GHz" and "3.6GHz" minimum clockspeed is a HWInfo bug, your chip is doing a lot of idle behind the scenes that only Ryzen Master can show.

Hate to steal your thunder.
Oh who am I kidding

P95 gives you an option of Smallest, Small, Large and Blend when you open it. Use either Smallest or Small for CPU stability (whichever gets hotter on your system) and Large for memory stability. Blend isn't great because P95's main selling point is that it's extremely intensive and generates a lot of heat; that makes it useful to see where the equilibrium of your cooling setup lies. When the temps basically stop climbing and settle around a particular point, you know how much your system can take. Blend would simply shift the load into something else and allow the part that's heated up to cool down again.

How long should I try P95 on smallest and small, large for each time?
been running smallest now for 15 minutes and CPU hottest was 81c and ram is between 40.5 and 43.8c
ran small for 15 minutes, highest CPU temp was 76c and ram was between 41.5 and 44.5c
ran large for 15 minutes, highest CPU was 76c and ram was between 42 and 45c

basically ran the first 5 sub tests of the 1st main test

Torture Test completed 5 tests in 16 minutes - 0 errors, 0 warnings.

 

[Zach_01]

oh agreed, i also remember those times, my first build was just after we moved here in '93

so UPS ?

Got my first in 2001 but had a lot of PC contact from middle 90s.... man... 20~33MHz CPU speed and 4MB of RAM . Golden words of history(1981): 640KB ought to be enough for anybody... right!!

Cant commend or recommend for a UPS. Not once used one and never had to. Even 20 years ago power loss in my area was rare and now is even better like once a year or not. And the quality (voltage) was always good enough without major flactuations.

 

[lorry]

Got my first in 2001 but had a lot of PC contact from middle 90s.... man... 20~33MHz CPU speed and 4MB of RAM . Golden words of history(1981): 640KB ought to be enough for anybody... right!!

Cant commend or recommend for a UPS. Not once used one and never had to. Even 20 years ago power loss in my area was rare and now is even better like once a year or not. And the quality (voltage) was always good enough without major flactuations.

My first ever was either the zx80 or 81,the one with rubber keys - somehow managed to poke to a peek only address (supposedly) and bricked it.

My first 'proper' was an Amstrad 286 that Originally only had 1MB of ram! I Finally upgraded it to 4MG when I installed win 3.0 on it, tried to run Quicken I think it was and sat watching the screen refresh line, BY line. Oh yeah And I upgraded the HDD to 40MB!

Anyways he was Only one letter wrong

G instead of K