Trying to understand Ryzen 3000 series boost speed variations

[tabascosauz]

Hate to steal your thunder.
Oh who am I kidding

View attachment 140999

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.

Well... I am surprised your single thread score is low. Is CB using an inferior core?


I used to run Smallest for like two hours at a time, haha. It can take a little longer than you'd expect for the temperature to stop climbing. But since no workload or other test matches the ridiculous heat output of Smallest, and Small is no longer viable due to the reduced multiplier (which your results show as well), I'm not sure if I would rely on Prime as much as I used to. IntelBurnTest is highly recommended for stability testing, it's quick to complete and it works pretty well without melting your rig like the old Small used to; it's just that IBT has never actually caught any CPU instabilities on my system, but P95 caught every one. YMMV

Now that I do think about it, the power grid was pretty safe when I was still at school in Bristol (will be going back again soon), didn't feel the need for a UPS over there at all. CyberPower and APC are the two big brands for consumer UPSes, but the popular models here don't seem to be available over there (the 1500VA simulated sine wave units from them, CP1500AVRLCD and BackUPS BR1500G, as well as the 1500VA pure sine wave units the CP1500PFCLCD and BackUPS Pro BR1500MS). I have the BR1500MS and as far as I can tell, CyberPower and APC are pretty much neck and neck in this market segment, maybe APC having the edge in build quality. For those two simulated sine and pure sine pairs, it's pretty much a "get whatever's cheaper/whatever you like better" affair.

I haven't heard of PowerCool, but the packaging and lack of a website suggests a rebranded generic product. It is a little cheaper than your typical 1350/1500VA from CP and APC, but I'm not sure I'd trust any of my equipment with that one. Just my $0.02. CP and APC (APC is the biggest) are large companies with support and warranty I think I can rely on, even if the equipment damage reimbursement guarantee inevitably requires you to jump thru the usual hoops. Also, buying a high profile and popular product means there's documentation on what and how to replace your batteries down the line.

 

[Waldorf]

Usually the cyberpower are quieter than apc, but those are a bit more powerful/efficient.

I would not recommend a simulated sine wave, active PFC psu's need a proper real sine output, or they'll take a hit and go out sooner.

The wattage is to make sure your not running above ups capabilities,
as it will just pop the fuse that you cant reset until below units max.
But your not getting a ups to run it of the battery (those are lead based and dont like longer discharges, similar to starter batteries in cars),
only so you can prevent crash/save your work before proper shutdown,
and to filter incoming power for noise (appliances in the house) and fluctuations on the grid that dont lead to brown/blackout.
Even when you have power that runs your stuff, doesn't mean its not outside the range the psu should have, nor does it mean its stable.
Even if you have numbers within lets say +/- 10 volt, its stress for a psu to constantly have to ramp up/down to make up for the changes and keep steady power going into the rig. Now add different draws during gaming whennloaf changes..

So far i have had not one psu die before the 5-7y warranty was over,
and i like it when it comes to bios/os updates, or even drives that dont get bricked because power issues.

If you can get one, around 1000VA should be fine.
Mine handles R5 3600 and oced 2080 and 32in moni plus other accessories just fine.

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

www.amazon.co.uk

If you can get any other brand with 1000VA or more and real sine out for less, buy it.

 

[tabascosauz]

Usually the cyberpower are quieter than apc, but those are a bit more powerful/efficient.

I would not recommend a simulated sine wave, active PFC psu's need a proper real sine output, or they'll take a hit and go out sooner.

The wattage is to make sure your not running above ups capabilities,
as it will just pop the fuse that you cant reset until below units max.
But your not getting a ups to run it of the battery (those are lead based and dont like longer discharges, similar to starter batteries in cars),
only so you can prevent crash/save your work before proper shutdown,
and to filter incoming power for noise (appliances in the house) and fluctuations on the grid that dont lead to brown/blackout.
Even when you have power that runs your stuff, doesn't mean its not outside the range the psu should have, nor does it mean its stable.
Even if you have numbers within lets say +/- 10 volt, its stress for a psu to constantly have to ramp up/down to make up for the changes and keep steady power going into the rig. Now add different draws during gaming whennloaf changes..

So far i have had not one psu die before the 5-7y warranty was over,
and i like it when it comes to bios/os updates, or even drives that dont get bricked because power issues.

If you can get one, around 1000VA should be fine.
Mine handles R5 3600 and oced 2080 and 32in moni plus other accessories just fine.

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

www.amazon.co.uk

If you can get any other brand with 1000VA or more and real sine out for less, buy it.

The BackUPS Pro noise level was an early batch thing. Mine's as quiet as a church mouse on battery power and plugged in. Now, if we were pushing 800W of power draw on either the CP1500PFCLCD or BR1500MS, it could be a different story...

Not that it matters, since it looks like Cyberpower is the only vendor with pure sine wave units in the UK. APC UK lacks the BR-MS line of BackUPS Pro, which is the only consumer line with sine wave. There are a lot of simulated/square wave BackUPS from APC, some of which are pretty affordable, but again, not sinewave.

You can get sinewave from APC in the SmartUPS line, but that's the commercially oriented heavy duty line that'll cost up upwards of 500 quid for a 1500VA unit, and will probably break your back when you try to lift it on your own.

Also, what the hell is up with the prices over there? The sine wave units are all priced like they've got gold batteries. The CP1500PFCLCD costs $300 here and the BR1500MS $275; on Amazon.co.uk, the CP1500PFCLCD is £266, the equivalent of $456 Canuckistan pesos.

 

[Waldorf]

sorry, should have said fan noise (under any load), but yeah not day&night difference.

The smartups is really nice, tho by the time its something i can use for more than just the pc, its close to a (small house) backup generator.

Yeah, i would check local pc shops, maybe open box from amazon..

 

[lorry]

Well... I am surprised your single thread score is low. Is CB using an inferior core?
View attachment 141013

I used to run Smallest for like two hours at a time, haha. It can take a little longer than you'd expect for the temperature to stop climbing. But since no workload or other test matches the ridiculous heat output of Smallest, and Small is no longer viable due to the reduced multiplier (which your results show as well), I'm not sure if I would rely on Prime as much as I used to. IntelBurnTest is highly recommended for stability testing, it's quick to complete and it works pretty well without melting your rig like the old Small used to; it's just that IBT has never actually caught any CPU instabilities on my system, but P95 caught every one. YMMV

Now that I do think about it, the power grid was pretty safe when I was still at school in Bristol (will be going back again soon), didn't feel the need for a UPS over there at all. CyberPower and APC are the two big brands for consumer UPSes, but the popular models here don't seem to be available over there (the 1500VA simulated sine wave units from them, CP1500AVRLCD and BackUPS BR1500G, as well as the 1500VA pure sine wave units the CP1500PFCLCD and BackUPS Pro BR1500MS). I have the BR1500MS and as far as I can tell, CyberPower and APC are pretty much neck and neck in this market segment, maybe APC having the edge in build quality. For those two simulated sine and pure sine pairs, it's pretty much a "get whatever's cheaper/whatever you like better" affair.

I haven't heard of PowerCool, but the packaging and lack of a website suggests a rebranded generic product. It is a little cheaper than your typical 1350/1500VA from CP and APC, but I'm not sure I'd trust any of my equipment with that one. Just my $0.02. CP and APC (APC is the biggest) are large companies with support and warranty I think I can rely on, even if the equipment damage reimbursement guarantee inevitably requires you to jump thru the usual hoops. Also, buying a high profile and popular product means there's documentation on what and how to replace your batteries down the line.

I have no idea, but my R20 results don't seem to be worth a damn as they vary by such a vast amount.
I can do one run and get around say 6900, do another right away and then get around 7200, so Why 300 point difference? Also that 3950x score I posted a couple of pages back was 'only' 519 and from what I can see people's R29 scores seem to be all over the place (see the thread 'post your R20 scores' on here, or the HWbot scores).

I knew that my P95 runs were too short but this was all I could do given the time I had left before I usually went to bed, so decided on a short 15 minute burst on each, I'll run a longer test today on smallest and largest

I will have a think and If I decide to get one I'll likely go with a APC but it's not a priority so much as my power supply seems pretty consistant nowadays.

Thing that Does bug me somewhat though and makes me unsure of my BIOS settings is that in HWinfo I have never seen the 'effective' go above roughly 4300 - do they not show a single core on that?
The worst chiplet never reaches more than 4150 as well

Usually the cyberpower are quieter than apc, but those are a bit more powerful/efficient.

I would not recommend a simulated sine wave, active PFC psu's need a proper real sine output, or they'll take a hit and go out sooner.

The wattage is to make sure your not running above ups capabilities,
as it will just pop the fuse that you cant reset until below units max.
But your not getting a ups to run it of the battery (those are lead based and dont like longer discharges, similar to starter batteries in cars),
only so you can prevent crash/save your work before proper shutdown,
and to filter incoming power for noise (appliances in the house) and fluctuations on the grid that dont lead to brown/blackout.
Even when you have power that runs your stuff, doesn't mean its not outside the range the psu should have, nor does it mean its stable.
Even if you have numbers within lets say +/- 10 volt, its stress for a psu to constantly have to ramp up/down to make up for the changes and keep steady power going into the rig. Now add different draws during gaming whennloaf changes..

So far i have had not one psu die before the 5-7y warranty was over,
and i like it when it comes to bios/os updates, or even drives that dont get bricked because power issues.

If you can get one, around 1000VA should be fine.
Mine handles R5 3600 and oced 2080 and 32in moni plus other accessories just fine.

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

www.amazon.co.uk

If you can get any other brand with 1000VA or more and real sine out for less, buy it.

How about this one?

CyberPower 1200VA/720W Tower Line-Interactive UPS IEC Green Power

Buy from Scan - 1200VA CyberPower VALUE1200EILCD Tower UPS Line-Interactive IEC Green Power 6 Outputs

www.scan.co.uk

 

[tabascosauz]

I have no idea, but my R20 results don't seem to be worth a damn as they vary by such a vast amount.
I can do one run and get around say 6900, do another right away and then get around 7200, so Why 300 point difference? Also that 3950x score I posted a couple of pages back was 'only' 519 and from what I can see people's R29 scores seem to be all over the place (see the thread 'post your R20 scores' on here, or the HWbot scores).

I knew that my P95 runs were too short but this was all I could do given the time I had left before I usually went to bed, so decided on a short 15 minute burst on each, I'll run a longer test today on smallest and largest

I will have a think and If I decide to get one I'll likely go with a APC but it's not a priority so much as my power supply seems pretty consistant nowadays.

Thing that Does bug me somewhat though and makes me unsure of my BIOS settings is that in HWinfo I have never seen the 'effective' go above roughly 4300 - do they not show a single core on that?
The worst chiplet never reaches more than 4150 as well

How about this one?

CyberPower 1200VA/720W Tower Line-Interactive UPS IEC Green Power

Buy from Scan - 1200VA CyberPower VALUE1200EILCD Tower UPS Line-Interactive IEC Green Power 6 Outputs

www.scan.co.uk

Yeah, as long as you don't have performance issues and get roughly the right performance out of your 2070 Super, not much point in running R20 all the time. Luck of the draw. Mine topped out at 4958 and I've yet to break that, even with unchanged settings. But my scores have never dipped below 4900, so perhaps the larger difference for you is the larger chip making for wilder boost or things running in the background.

Effective clock really depends on what you happen to be doing on a certain day. I've been playing a bit of BF3 this week, trip down memory lane, and it's really CPU limited and limited to like 2 cores. My GPU is barely even working at all, doesn't break 50 degrees. Yet while I only usually see about 4200MHz of effective clock:



If you're running a lot of the cores a lot of the time, clocks will probably be lower. A really CPU-limited older game might give you something more like this, instead. If you have HWInfo in the background while you bench CB R20 or CPU-Z single-thread, it'll probably boost higher.

That 1200EILCD is probably part of the value line, which explains the blocky design departure from CyberPower's usual fare and simulated sinewave. The thing about simulated sine is that there are a lot of people who get by with no problems at all, but you're essentially playing the lottery with whether your PSU will take a liking to it or not when you buy simulated sine. If it works, great. If it doesn't work, well, you'll find out after the first blackout that you just bought an expensive paperweight.

It's really hard to find professional reviews for these obscure, low-end units. If Amazon reviews are anything to go by...I dunno about this one.

 

[lorry]

Seen these two ?

Rackmount | Uninterruptible Power Supply Rackmounts | Ebuyer.com

Browse our range of Rackmount UPS Uninterruptible Power Supply Units. Protect data loss and more with our Rackmount systems at the best prices with fast delivery from Ebuyer!

www.ebuyer.com

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

www.amazon.co.uk

 

[tabascosauz]

Seen these two ?

Rackmount | Uninterruptible Power Supply Rackmounts | Ebuyer.com

Browse our range of Rackmount UPS Uninterruptible Power Supply Units. Protect data loss and more with our Rackmount systems at the best prices with fast delivery from Ebuyer!

www.ebuyer.com

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

CyberPower CP900EPFCLCD-UK PFC Series, 900VA/540W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible, Tower: Amazon.co.uk: Computers & Accessories

www.amazon.co.uk

Yeah, that's the sinewave line I was talking about when I was referring to the 266 quid CP1500PFCLCD. It's quite popular here in North America. I'm not sure I'd want to go as low as 900VA; it's nice to have some room to plug in some other stuff too without worrying about running out of wattage when gaming.

The 900VA is 540W; I regularly pull upwards of 300W of system+monitor load on my 3700X and 2060S, whereas yours draws more power in both areas. I'm guessing you have more fans that draw a bit more power than me, as well.

Also, a nice thing about these UPSes is that if CyberPower is anything like APC, you connect a USB cable to your computer and it sends information over ACPI so you can monitor everything about it in HWInfo. As long as it conforms to ACPI, it shows up in Windows as a battery and HWInfo detects it as a UPS.

 

[Waldorf]

the 1200/720 is just simulated output.

like apc, cyberpower only has two true sine out lineups for consumer, the CPxxxxPFCLCD series is the cheaper one.

i never meant the output of your psu, but the power it gets fed.
the grid isnt that clean, appliances in your home do the rest, add voltage fluctuations and voila.
i had 5-10 registers down at the store i did cover IT for a year, fixed by replacing the batteries for all UPS (no AVR, so they needed the battery to boost sub 120V).

Also never seen more stable output from psu for "fixed" voltages (IF, i can see a change every couple seconds).

In short, i will not run anything above 200$ without it (even the tv/dvr/router get a APC 650VA with AVR),
and same for units i build for others.

So dont starve yourself for 2 month to get it, but i recommend one with a decent build like yours.
Any chance of buying it in europe (trip/thru family etc)

 

[lorry]

This one is £190

CyberPower Intelligent 1300VA LCD PFC Series UPS | Ebuyer.com

Get a great deal on a CyberPower Intelligent 1300VA LCD PFC Series UPS as well as thousands of products at Ebuyer!

www.ebuyer.com

 

[tabascosauz]

This one is £190

CyberPower Intelligent 1300VA LCD PFC Series UPS | Ebuyer.com

Get a great deal on a CyberPower Intelligent 1300VA LCD PFC Series UPS as well as thousands of products at Ebuyer!

www.ebuyer.com

Looks like a decent deal, especially compared to price of the Value line one. For the APC counterparts, the BR1350MS has cheaper but pretty good batteries that possibly rivals the BR1500MS in runtime, for an appreciably lower price tag. It could be a similar story with the CyberPower 1350 and 1500PFCLCDs, which would be pleasant if that was the case.

 

[Waldorf]

the "bigger" units usually just have a bigger battery inside (1000/1300 and 1300/1500), and less outlets.
compare the specs and if its only the battery, get the cheaper unit and once you need to replace the battery (+5y),
just order the bigger one.

The 900VA should be fine, your hardware isnt drawing more than 500w,
and its not supposed to run on battery anyway, so the 1 min (if power goes out) it takes to close the game and shutdown pc is gonna be fine.
when you run a UPS connected to the pc with usb (data), you will get additional power options (cpu/gpu/pcie etc) that you can use to minimize power load.

e.g. if i have a blackout i wont keep playing anyway, so having lower game performance for a minute (because of energy settings when the rig is on battery),
might lose you a round/game, but will limit stress on the ups (no matter what size).

its like a flashlights, you dont use em all the time (to replace regular lights)

What's the Difference Between Pure Sine Wave and Simulated Sine Wave?

Shopping for a UPS? Two types exist—and their respective output waveforms do have an effect on the equipment they're backing up.

blog.tripplite.com

so no tripplite units that say its pwm output...

PFC Sinewave Archives

CyberPower manufactures high-quality PFC Sinewave products for consumers and IT professionals.

www.cyberpowersystems.com

and if you need to hook up other stuff, i would rather get multiple small units,
which makes it cheaper (simulated sine out is ok), and anything thats "dumb" tech/cheap to replace, doesnt even need the AVR feature.

Tripp Lite AVRX550U UPS

uk.pcpartpicker.com

Tripp Lite AVRX750U UPS

uk.pcpartpicker.com

 

[lorry]

The only APC that I can see in the UK is this one

Backup Power | UPS | Generators & More

UK #1 for UPS | Generators | PDU | Comms Racks & more www.escpower.co.uk/ Tripp Lite : Riello : APC : Clarke : CertaUPS : Borri : CE&T Power : All-Rack For all your power protection, comms rack enclosure and power distribution needs.

www.escpower.co.uk

 

[tabascosauz]

The only APC that I can see in the UK is this one

Backup Power | UPS | Generators & More

UK #1 for UPS | Generators | PDU | Comms Racks & more www.escpower.co.uk/ Tripp Lite : Riello : APC : Clarke : CertaUPS : Borri : CE&T Power : All-Rack For all your power protection, comms rack enclosure and power distribution needs.

www.escpower.co.uk

That one's still simulated sine. Unlike PCs, which just take whatever cable with the right plug on one end and the C14 connector on the other end, UPSes have the lead coming out of the unit itself. Might be why APC decided to run completely different product lines for different regions.

The BackUPS Pros are split into the MI and SI lines for the 230V market over there. MI is the cheap stepped wave units, SI is the counterpart to the 110V BR-MS units here (even look the same). Except the SIs are unfathomable expensive, like 500 quid. That's SmartUPS pricing......and that's by far the better product.

So just pick out a CyberPower unit you like.

 

[lorry]

the "bigger" units usually just have a bigger battery inside (1000/1300 and 1300/1500), and less outlets.
compare the specs and if its only the battery, get the cheaper unit and once you need to replace the battery (+5y),
just order the bigger one.

The 900VA should be fine, your hardware isnt drawing more than 500w,
and its not supposed to run on battery anyway, so the 1 min (if power goes out) it takes to close the game and shutdown pc is gonna be fine.
when you run a UPS connected to the pc with usb (data), you will get additional power options (cpu/gpu/pcie etc) that you can use to minimize power load.

e.g. if i have a blackout i wont keep playing anyway, so having lower game performance for a minute (because of energy settings when the rig is on battery),
might lose you a round/game, but will limit stress on the ups (no matter what size).

its like a flashlights, you dont use em all the time (to replace regular lights)

What's the Difference Between Pure Sine Wave and Simulated Sine Wave?

Shopping for a UPS? Two types exist—and their respective output waveforms do have an effect on the equipment they're backing up.

blog.tripplite.com

so no tripplite units that say its pwm output...

PFC Sinewave Archives

CyberPower manufactures high-quality PFC Sinewave products for consumers and IT professionals.

www.cyberpowersystems.com

and if you need to hook up other stuff, i would rather get multiple small units,
which makes it cheaper (simulated sine out is ok), and anything thats "dumb" tech/cheap to replace, doesnt even need the AVR feature.

Tripp Lite AVRX550U UPS

uk.pcpartpicker.com

Tripp Lite AVRX750U UPS

uk.pcpartpicker.com

"It’s important to understand that both types of UPS system produce true sine wave output more than 99% of the time. It is only during a power failure, when the UPS is producing power from its battery reserves, the output waveform is a concern."

if it is only during a power outage, how much damage is likely to be done in the one minute that it takes to shut down your PC?

As in how bad could this one be for instance -
CyberPower CP1300EPFCLCD-UK PFC Series, 1300VA/780W, 6 Outlets (4 IEC + 2 UK), AVR, Active PFC compatible

CyberPower CP1300EPFCLCD-UK PFC Series, 1300VA: Amazon.co.uk: Electronics

CyberPower CP1300EPFCLCD-UK PFC Series, 1300VA: Amazon.co.uk: Electronics

www.amazon.co.uk

Yeah, as long as you don't have performance issues and get roughly the right performance out of your 2070 Super, not much point in running R20 all the time. Luck of the draw. Mine topped out at 4958 and I've yet to break that, even with unchanged settings. But my scores have never dipped below 4900, so perhaps the larger difference for you is the larger chip making for wilder boost or things running in the background.

Effective clock really depends on what you happen to be doing on a certain day. I've been playing a bit of BF3 this week, trip down memory lane, and it's really CPU limited and limited to like 2 cores. My GPU is barely even working at all, doesn't break 50 degrees. Yet while I only usually see about 4200MHz of effective clock:

View attachment 141035

If you're running a lot of the cores a lot of the time, clocks will probably be lower. A really CPU-limited older game might give you something more like this, instead. If you have HWInfo in the background while you bench CB R20 or CPU-Z single-thread, it'll probably boost higher.

That is my point though - My scores vary by Far more than that, there is often up to a 500 point variance from min to max!
As in lowest score will be 68** and highest will be 73** ?

That looks to be one hell of a range to me, which is Why I wondered if my settings were correct?

Also, the Max effective clock seen today (and every day) was 4299 with 'effective clock' but 4575 with 'perf' ?



-----------------------------------------------------------------------------------------

Just finished a 2-hour run of smallest in P95 -

highest CPU temp was 82c with it often sitting in the 60s, ram ranged from 40 - 43c highest readings
About to start the memory one now

 

[Waldorf]

started with a corsair 750 that died on me within 2y (5y warranty) and i never stressed it (max 350 w).
decided to sell the ups (only pwm output) and go for the cyber power PFC-LCD unit with true sine out.
maintenance cost is the same between them, as they use the same battery (around 20-30$ (at least over here), and last about 5y),

since i started using true sine out (2010), all psus since then (about 10) are still running (5y+),
incl the last one from 2015 i just sold.
so you pay a bit more up front, but on long run possibly less on power bill (not much i guess),
and most likely longer lasting HW, that could make up for it over time.


i just like the fact that the psu has less stress to keep steady power going out to the pc,
the ups filters crap out of the signal, works as surge protector, and can help keeping input voltage constant
(make sure to set (any) unit to the narrower range, so voltage outside that range gets boosted/discarded. e.g. 210-230 instead of stock/wider range of 200-240V).
and i never worry about my hw getting fried (+100K insurance for surge) and no chance of bricking board/gpu/ODD from power outage while updating FW.

its like most other stuff.
when you buy a +80K car, you dont fill it with cheap gas ore buy used tires.
to me, a decent rig that is being used for more than surfing/streaming (e.g. sees load), deserves proper power.
at least on the long run i recommend anyone to get one.

hey, i even watched tv on it for a while (dtv box and moni) when power was out all day and i was tired of reading a book XD

 

[lorry]

[Jan 2 13:04] Torture Test completed 52 tests in 2 hours, 3 minutes - 0 errors, 0 warnings.

Ram test ran for 2 hours, but this time I adjusted the RPM/temp settings of All the fans (2 x 200mm intake, 140mm exhaust and both CPU fans), so that under stress they were at full speed, mostly to A) see just how loud they were and B) what effect that would have. It Was noisy but not massively so, did help to reduce the temps a decent amount on the ram though (likely the higher airflow from the two large intake fans flowing over the ram).
Result - ram temps never reached 40c and CPU mostly remained in the 60 - 75c range, so that's ok.

Interestingly though, even their Large test does not make anyway near to most of my ram, in fact not even 50% of it

started with a corsair 750 that died on me within 2y (5y warranty) and i never stressed it (max 350 w).
decided to sell the ups (only pwm output) and go for the cyber power PFC-LCD unit with true sine out.
maintenance cost is the same between them, as they use the same battery (around 20-30$ (at least over here), and last about 5y),

since i started using true sine out (2010), all psus since then (about 10) are still running (5y+),
incl the last one from 2015 i just sold.
so you pay a bit more up front, but on long run possibly less on power bill (not much i guess),
and most likely longer lasting HW, that could make up for it over time.


i just like the fact that the psu has less stress to keep steady power going out to the pc,
the ups filters crap out of the signal, works as surge protector, and can help keeping input voltage constant
(make sure to set (any) unit to the narrower range, so voltage outside that range gets boosted/discarded. e.g. 210-230 instead of stock/wider range of 200-240V).
and i never worry about my hw getting fried (+100K insurance for surge) and no chance of bricking board/gpu/ODD from power outage while updating FW.

its like most other stuff.
when you buy a +80K car, you dont fill it with cheap gas ore buy used tires.
to me, a decent rig that is being used for more than surfing/streaming (e.g. sees load), deserves proper power.
at least on the long run i recommend anyone to get one.

hey, i even watched tv on it for a while (dtv box and moni) when power was out all day and i was tired of reading a book XD

looks like This one is my only realistic option then and that's going to take a good few months to realistically get the money together for it

CyberPower Intelligent 1300VA LCD PFC Series UPS | Ebuyer.com

Get a great deal on a CyberPower Intelligent 1300VA LCD PFC Series UPS as well as thousands of products at Ebuyer!

www.ebuyer.com

------------------------------------------------------------------------------------------------------------------------------------------------------------------


On a separate note I have a watts meter now, only thing is that unless mine is faulty or the backlight is a setting that I haven't found (have messaged the seller but don't hold out too much hope (Chinese seller) ).

UK Plug Energy Monitor Power Meter Electricity Wattage Watt Kwh Analyzer Socket | eBay

Find many great new & used options and get the best deals for UK Plug Energy Monitor Power Meter Electricity Wattage Watt Kwh Analyzer Socket at the best online prices at eBay! Free delivery for many products!

www.ebay.co.uk

----------------------------------------------------------------------------------------------

Geekbench 5.1 is out as well

 

[Waldorf]

yeah those clones usually dont have one, glad i ordered thru amazon and just returned it to get a proper one.

 

[lorry]

yeah those clones usually dont have one, glad i ordered thru amazon and just returned it to get a proper one.

Well unless this seller replies I will be returning it as well (paid through PayPal and never had any issues when returning that way).

yeah those clones usually dont have one, glad i ordered thru amazon and just returned it to get a proper one.

yeah, there is this one on Amazon that specifies backlit in the description

Nevsetpo Power Meter UK Plug Power Monitor Watts Meter Electricity Usage Cost Meter Monitor Plug : Amazon.co.uk: DIY & Tools

Shop Nevsetpo Power Meter UK Plug Power Monitor Watts Meter Electricity Usage Cost Meter Monitor Plug. Free delivery on eligible orders of £20 or more.

www.amazon.co.uk

See what happens now

 

[Waldorf]

Not sure if you already have an amazon account, but might be a good idea.
Depending on item, it can be cheaper/faster/easier to order stuff thru prime,
and with the monthly membership you just cancel prime afterwards.
I usually do this and save a little during month i dont use it (vs yearlong),
and every now and then they will offer trials from 3-14 days for 1.99-7.99.

Just the fact i got an open box/used liquid cooled 2080 for 620$,
a little more than a new 2070S (aircooled) would have cost and a lot less than normal price (750-800),was worth doing a monthof prime for 13$.
And 2 day shipping is nice as well..

 

[lorry]

Not sure if you already have an amazon account, but might be a good idea.
Depending on item, it can be cheaper/faster/easier to order stuff thru prime,
and with the monthly membership you just cancel prime afterwards.
I usually do this and save a little during month i dont use it (vs yearlong),
and every now and then they will offer trials from 3-14 days for 1.99-7.99.

Just the fact i got an open box/used liquid cooled 2080 for 620$,
a little more than a new 2070S (aircooled) would have cost and a lot less than normal price (750-800),was worth doing a monthof prime for 13$.
And 2 day shipping is nice as well..

Had prime for years, just that money is tight atm and the one on Ebay looked exactly the same but cheaper, I went with them. Most times that works out, this time it didn't.
I Always check for the best price, which Always includes postage (so if something is a bit dearer but free postage say, I go with the overall price) and for PC parts, that includes Amazon, eBay, Scan uk, Box Ebuyer, etc etc.

 

[Waldorf]

You know pcpartpicker.com, right?

 

[lorry]

You know pcpartpicker.com, right?

First build in two decades

Check out ldwilliams's completed build on PCPartPicker! Ryzen 9 3900X 3.8 GHz 12-Core, GeForce RTX 2070 SUPER 8 GB AORUS, MasterCase H500M ATX Mid Tower, ...

uk.pcpartpicker.com

----------------------------------------------------------------------------------------------------------------------------------------


----------------------------
IntelBurnTest v2.54
Created by AgentGOD
----------------------------

Processor: AMD Ryzen 9 3900X 12-Core Processor
Clock Speed: 3.01 GHz
Active Physical Cores: 24
Total System Memory: 32713 MB

Stress Level: Standard (1024 MB)
Testing started on 02/01/2020 9:12:48 PM
Time (s) Speed (GFlops) Result
[21:12:59] 6.713 133.1644 3.393900e-002
[21:13:11] 6.857 130.3562 3.393900e-002
[21:13:22] 6.770 132.0377 3.393900e-002
[21:13:34] 6.830 130.8688 3.393900e-002
[21:13:45] 6.793 131.5922 3.393900e-002
[21:13:57] 7.007 127.5776 3.393900e-002
[21:14:08] 6.909 129.3829 3.393900e-002
[21:14:19] 6.734 132.7389 3.393900e-002
[21:14:31] 6.738 132.6577 3.393900e-002
[21:14:42] 6.844 130.6082 3.393900e-002
Testing ended on 02/01/2020 9:14:42 PM
Test Result: Success.
----------------------------

----------------------------------------------------------------------------------------------------------------------------------------


----------------------------
IntelBurnTest v2.54
Created by AgentGOD
----------------------------

Processor: AMD Ryzen 9 3900X 12-Core Processor
Clock Speed: 3.01 GHz
Active Physical Cores: 24
Total System Memory: 32713 MB

Stress Level: High (2048 MB)
Testing started on 02/01/2020 9:16:29 PM
Time (s) Speed (GFlops) Result
[21:16:58] 19.101 138.3490 3.125684e-002
[21:17:27] 19.171 137.8381 3.125684e-002
[21:17:56] 19.340 136.6361 3.125684e-002
[21:18:24] 19.258 137.2172 3.125684e-002
[21:18:53] 19.154 137.9606 3.125684e-002
[21:19:21] 19.203 137.6088 3.125684e-002
[21:19:50] 19.282 137.0483 3.125684e-002
[21:20:19] 19.264 137.1757 3.125684e-002
[21:20:47] 19.223 137.4690 3.125684e-002
[21:21:16] 19.165 137.8835 3.125684e-002
Testing ended on 02/01/2020 9:21:16 PM
Test Result: Success.
----------------------------

----------------------------------------------------------------------------------------------------------------------------------------
IntelBurnTest v2.54
Created by AgentGOD
----------------------------

Processor: AMD Ryzen 9 3900X 12-Core Processor
Clock Speed: 3.01 GHz
Active Physical Cores: 24
Total System Memory: 32713 MB

Stress Level: Very High (4096 MB)
Testing started on 02/01/2020 9:23:31 PM
Time (s) Speed (GFlops) Result
[21:24:45] 54.358 141.7952 3.197914e-002
[21:26:00] 55.283 139.4231 3.197914e-002
[21:27:14] 55.277 139.4376 3.197914e-002
[21:28:28] 55.231 139.5551 3.197914e-002
[21:29:43] 55.269 139.4581 3.197914e-002
[21:30:57] 55.068 139.9675 3.197914e-002
[21:32:11] 55.491 138.8998 3.197914e-002
[21:33:26] 55.287 139.4138 3.197914e-002
[21:34:40] 55.344 139.2693 3.197914e-002
[21:35:54] 55.140 139.7837 3.197914e-002
Testing ended on 02/01/2020 9:35:55 PM
Test Result: Success.
------------------------------------------------------------------------------------------------------------------------------------------------------------

Couldn't do maximum ?

At times the available ram was Down to 2MB! and the virtual memory usage was up to 49,985MB!!!

Hmm, seems I dint run it for long enough maybe. Just seen a report of each run taking 15 minutes plus, I stopped it after about 12 mins

Oh and cpu temp never went above 71c and ran was quite cool, 33.5 - 34.5c

Definitive guide to configuring the Ryzen 3900X

It has taken me this long to write this guide because I don't get free stuff sent to me so I had to wait until I was able to get hands on experience with this CPU and its two chiplet design. So I am sorry if you have been having a bad experience trying to make sense of what "influencers" who do...

community.amd.com

 

[Zach_01]

Thing that Does bug me somewhat though and makes me unsure of my BIOS settings is that in HWinfo I have never seen the 'effective' go above roughly 4300 - do they not show a single core on that?
The worst chiplet never reaches more than 4150 as well

Efffective clock is not for single core boost/clock. It contains the sleeping states of the cores, so the reported clock has such a low value. Think of it as an average between active and sleeping state.

FEATURE - Effective clock vs instant (discrete) clock

It has become a common practice for several years to report instant (discrete) clock values for CPUs. This method is based on knowledge of the actual bus clock (BCLK) and sampling of core ratios at specific time points. The resulting clock is then a simple result of ratio * BCLK. Such approach...

www.hwinfo.com

Another topic...
Have some premature (not alot testing) findings while playing around with EDC from UEFI that may interest you.
Remember a few posts back I was giving a (negative) offset to vcore so that overall power draw (PPT) and temp of CPU would drop and thus having (in theory) more headroom for clocks. That was not the case for my CPU as this was decreasing eventually clocks and performance, due to poor silicon quality that cannot sustain max clocks with direct under voltage. This offset is applied across the entire range of the clock (single and all core). Tho... the power effectiveness of the CPU was increased. I had decreased voltage by ~2% and lost clock/performance by ~1%.
This could work better with a higher silicon quality CPU like the 3900X.

So I decided to take another route by decreasing the CPU's current (EDC/Amps) and not voltage directly to control max power draw.
My default max values goes like this:
PPT: 88 Watt
TDC: 60 Amps
EDC: 90 Amps

When running R20 all core, those limits was never reached except PPT, so:
PPT: 87.7 Watt
TDC: ~49 Amps
EDC: ~79 Amps
Max temp: 63C

Into UEFI I set the EDC limit to 76A and run R20:
PPT: 87.8 Watt
TDC: ~49 Amps
EDC: 76 Amps
Max temp: 63C

Into UEFI I set the EDC limit to 73A and run R20:
PPT: 88 Watt
TDC: ~49 Amps
EDC: 73 Amps
Max temp: 63C

This indicates that in order to still having the same PPT and temp, while power current (EDC) has dropped... something else must have gotten up! The only thing that could have gotten up is the clock...
Even though I didnt see a clock increase visually, I did see some increase in R20 scores for the few runs I did. The "drawback" in the EDC adjustment is that only affects max power draw, so only max all core clocks/boost... unlike the voltage offset that affects the entire range of clocks.
I will come back with this, but feel free to do some testing yourself (voltage offset or EDC). I will even test with a combination of the two (negative voltage offset + limited EDC).

 

[lorry]

Efffective clock is not for single core boost/clock. It contains the sleeping states of the cores, so the reported clock has such a low value. Think of it as an average between active and sleeping state.

FEATURE - Effective clock vs instant (discrete) clock

It has become a common practice for several years to report instant (discrete) clock values for CPUs. This method is based on knowledge of the actual bus clock (BCLK) and sampling of core ratios at specific time points. The resulting clock is then a simple result of ratio * BCLK. Such approach...

www.hwinfo.com

Another topic...
Have some premature (not alot testing) findings while playing around with EDC from UEFI that may interest you.
Remember a few posts back I was giving a (negative) offset to vcore so that overall power draw (PPT) and temp of CPU would drop and thus having (in theory) more headroom for clocks. That was not the case for my CPU as this was decreasing eventually clocks and performance, due to poor silicon quality that cannot sustain max clocks with direct under voltage. This offset is applied across the entire range of the clock (single and all core). Tho... the power effectiveness of the CPU was increased. I had decreased voltage by ~2% and lost clock/performance by ~1%.
This could work better with a higher silicon quality CPU like the 3900X.

So I decided to take another route by decreasing the CPU's current (EDC/Amps) and not voltage directly to control max power draw.
My default max values goes like this:
PPT: 88 Watt
TDC: 60 Amps
EDC: 90 Amps

When running R20 all core, those limits was never reached except PPT, so:
PPT: 87.7 Watt
TDC: ~49 Amps
EDC: ~79 Amps
Max temp: 63C

Into UEFI I set the EDC limit to 76A and run R20:
PPT: 87.8 Watt
TDC: ~49 Amps
EDC: 76 Amps
Max temp: 63C

Into UEFI I set the EDC limit to 73A and run R20:
PPT: 88 Watt
TDC: ~49 Amps
EDC: 73 Amps
Max temp: 63C

This indicates that in order to still having the same PPT and temp, while power current (EDC) has dropped... something else must have gotten up! The only thing that could have gotten up is the clock...
Even though I didnt see a clock increase visually, I did see some increase in R20 scores for the few runs I did. The "drawback" in the EDC adjustment is that only affects max power draw, so only max all core clocks/boost... unlike the voltage offset that affects the entire range of clocks.
I will come back with this, but feel free to do some testing yourself (voltage offset or EDC). I will even test with a combination of the two (negative voltage offset + limited EDC).

That does look to be interesting but I feel that I would need to know how far to go in undervolting and/or EDC - for instance do you reduce or increase EDC?

I just ran CB20 and here are my max limits, you can see that I do reach 100% on EDC and close to the limits on the other two




Also the discusion thread in this post is interesting, it talks about FIT, setting voltage limits etc, might be worth a read?

Definitive guide to configuring the Ryzen 3900X

It has taken me this long to write this guide because I don't get free stuff sent to me so I had to wait until I was able to get hands on experience with this CPU and its two chiplet design. So I am sorry if you have been having a bad experience trying to make sense of what "influencers" who do...

community.amd.com