• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

PBO / XFR 3000 series - Definitive values

Brizzles said:
Nope, does the opposite for me.

Screenshots show the clock speeds taken from a 2 minute run ingame ( same environment in game for both tests ) ;

Core Voltage - AUTO with XFR / PBO2 - DISABLED ( Average clock speed 4182MHz )

ukLPnG9.png


Core Voltage - AUTO with XFR AUTO / PBO2 ENABLED ( Average clock speed 4116MHz )

Ah46LRr.png


The minimum , maximum and average are all higher with XFR and PBO2 Disabled
Click to expand...

How can it be the same if total CPU utilisation is >25% higher (10.4% vs 8.0%).
 
muon said:
How can it be the same if total CPU utilisation is >25% higher (10.4% vs 8.0%).
Click to expand...

not sure if it's the same thing or not, but on my own overclocking testing I've found my 3950x shows higher % usage at stock yet has lower performance than the overclock which shows lower usage %. The difference is very small, we're talking like 1-2% but it's there - it's very weird
 
muon said:
How can it be the same if total CPU utilisation is >25% higher (10.4% vs 8.0%).
Click to expand...

True, by same i meant it was the same area in the game i had to hand - Just a quick and easy test.

So, i've controlled it with a random benchmark - Resident Evil 6 Benchmark @ 1920 x 1080 ;


Core Voltage AUTO
XFR DISABLED
PBO2 DISABLED

Average clock speed 4052
Average Boost/Max 4412

Res Evil 6 Bench Scores 8279 / 8159 / 8148
wlHUiGc.png



Core Voltage AUTO
XFR AUTO
PBO2 ENABLED

Average clock speed 4015
Average Boost/Max clock 4367

Res Evil 6 Bench Scores 8665/8600/8548
uKsKL7c.png


The results pan out the same as the first 'quick' test

XFR and PBO2 Disabled produces a higher average clock speed and higher boost//max clock speed than with them enabled.

But, just to throw a spanner in the works, despite having higher average and boost clocks it scores around 5% less in the benchmark.

Am i missing something here ???
 
Kevin Matthews said:
@Brizzles Ryzen master is there cores going to sleep?

Also what power plan you using and what is min processor set at?

With it disable your using your best 3cores @4.475 compared to 2 with it enabled.
Click to expand...

1 ) Don't use Ryzen Master anymore, just bios settings

2 ) Haven't set a minimum processor to my knowledge

3 ) 1usmus Ryzen Universal power plan ( including the BIOS recomendations )

I'm confused as to why lower average clock speeds yield a 5% higher bench mark score
 
What's your cooling like @Brizzles ?

Not disputing your results. They are what they are, it's just strange that when you give the CPU more headroom to perform it actually performs slower.

This would indicate that PBO/XFR isn't functioning properly because the very definition of it (if it had a definition) is raising the power limits to enable your CPU to boost higher and for longer.

Given the net result you are seeing is less performance then I'd say it isn't working to specification.

For me it definitely is worth it. + 200 points in Cinebench and a good 5% more flops in Intel Burn test.

In the end I have just left it enabled with auto scalar.
 
Brizzles said:
True, by same i meant it was the same area in the game i had to hand - Just a quick and easy test.

So, i've controlled it with a random benchmark - Resident Evil 6 Benchmark @ 1920 x 1080 ;


Core Voltage AUTO
XFR DISABLED
PBO2 DISABLED

Average clock speed 4052
Average Boost/Max 4412

Res Evil 6 Bench Scores 8279 / 8159 / 8148
wlHUiGc.png



Core Voltage AUTO
XFR AUTO
PBO2 ENABLED

Average clock speed 4015
Average Boost/Max clock 4367

Res Evil 6 Bench Scores 8665/8600/8548
uKsKL7c.png


The results pan out the same as the first 'quick' test

XFR and PBO2 Disabled produces a higher average clock speed and higher boost//max clock speed than with them enabled.

But, just to throw a spanner in the works, despite having higher average and boost clocks it scores around 5% less in the benchmark.

Am i missing something here ???
Click to expand...

There should be a section in Hwinfo with the true effective clock speeds. See what tgat averages at if you have time.

I always got higher 3dmark scores with PBO and then slightly higher again with PBO+AutoOC. We are only talking about 50-100mhz worth of improvement in total though.
 
Last edited:
Brizzles said:
1 ) Don't use Ryzen Master anymore, just bios settings

2 ) Haven't set a minimum processor to my knowledge

3 ) 1usmus Ryzen Universal power plan ( including the BIOS recomendations )

I'm confused as to why lower average clock speeds yield a 5% higher bench mark score
Click to expand...


Ryzen master good for temps and readings for cores don't use it for anything else.

What bios are you on your motherboard? I'm using the new power plan for 1.0.0.4 bios but I'm MSI. Use to 1usmus but didn't find it much difference on the 3600.
 
opethdisciple said:
What's your cooling like @Brizzles ?

Not disputing your results. They are what they are, it's just strange that when you give the CPU more headroom to perform it actually performs slower.

This would indicate that PBO/XFR isn't functioning properly because the very definition of it (if it had a definition) is raising the power limits to enable your CPU to boost higher and for longer.

Given the net result you are seeing is less performance then I'd say it isn't working to specification.

For me it definitely is worth it. + 200 points in Cinebench and a good 5% more flops in Intel Burn test.

In the end I have just left it enabled with auto scalar.
Click to expand...

Cooling is a Corsair 360 AIO, max temps ~70 degrees load

Was giving the results some thought later, and perhaps i'm looking in the wrong place with actual clock speeds.

I've been focussing on the 'Boost' part in Precion Boost Overdrive expecting clocks to 'boost' i.e be higher.

If PBO is not physically boosting clocks, but using higher cores for longer than lower cores then it would make sense that i got a 5% higher bench with lower average clocks.

My simple way of justifying it to myself would be :

core 1 4500
core 2 4250

Average clock speed 4375

If PBO was telling the chip to use core 1 90% of the time, same high/low/average cloch speed but higher performance
If PBO was telling the chip to use core 2 90% of the time, same high/low/average clock speed but lower performance

Sound right ?
 
muon said:
There should be a section in Hwinfo with the true effective clock speeds. See what tgat averages at if you have time.

I always got higher 3dmark scores with PBO and then slightly higher again with PBO+AutoOC. We are only talking about 50-100mhz worth of improvement in total though.
Click to expand...

Will do some more digging around later, will use your suggestion - Thanks
 
I found similar when playing around.

PE3
PBO Manual
PPT 0
TDC 0
EDC 1
Overdrive 200mhz

I get (upto):
4.675GHz on cores 0+1
4.625GHz on cores 2+3
4.600GHz on cores 4+5
4.400GHZ to 4.475GHz on cores 6 to 11

All core is 4.250GHz

i.e. it boosts lovely! However, despite this it nets just 7492 points in CB20 vs 7481 points I can score stock..

Which Is why i leave it stock.
 
Last edited:
PBO gives me another 100 points or so in Cinebench R20, but its just not worth the additional power consumption / heat generation for such a marginal gain. I just leave everything on Auto and it performs as advertised.
 
PBO Manual with:
PPT 0
TDC 0
EDC 1
Overdrive 200mhz

Actually runs quite a bit cooler for me.
 
Last edited:
Realised I'm not running the 1usmus power plan any more, I just remembered why, it cripples single core boost for some reason. So I actually use Ryzen High Performance with 99% min cpu state.
 
Yaayuh! said:
Realised I'm not running the 1usmus power plan any more, I just remembered why, it cripples single core boost for some reason. So I actually use Ryzen High Performance with 99% min cpu state.
Click to expand...

Actually, scrap that.

PBO itself borks single core for me. Clocks run under 3GHz when running CB20 single core test, snails pace.

I do not know why.

Back to stock, I knew having another go at tinkering would be a fruitless endeavour.
 
As a test I disabled PBO And I am hitting 100% limits in either PPT or EDC and I have lost 5 flops in Intel Burn test.

As far as I'm concerned for the CPU to be achieving what it's capable of it should never hit 100% of any of the three metrics.
 
You must log in or register to reply here.
Back
Top Bottom