Ryzen Owners Zen Garden

[Mussels]

not reducing temperatures of the CPU, reducing the wattage/heat fed into my PC

hot air from the CPU is fed into my case air, which cools my GPU AIO - so every watt i save there, has a payoff in lowering the GPU temps

 

[lexluthermiester]

not reducing temperatures of the CPU, reducing the wattage/heat fed into my PC

hot air from the CPU is fed into my case air, which cools my GPU AIO - so every watt i save there, has a payoff in lowering the GPU temps

Front mounted rad?

 

[TheLostSwede]

I'm not sure you should go into Ryzen 3000 with the intention of reducing temperatures, as you may find yourself in for an unpleasant surprise. Now, with performance and efficiency in mind, you most definitely should

So far, I really like what 1.0.0.3ABBA has done. Idle is stable on temperatures, and a firmware change now allows HWInfo to read CCD temp (the actual cores portion as opposed to the usual Tdie. Benchmarks aren't being favoured as the multipliers quickly back down from 43-44x to 40x, but I suspect that enabling PBO again would quickly solve that issue for benchmark warriors. Light benches like CPU-Z perform well. On the other hand, 44x is being achieved on roughly half of the cores, whereas ABB would top out at 43.5x on just the fastest core.

End result is that gaming stays quiet around 55-60c with max boost on the cores that need it, while benchmarking is reasonable at 60-70 up to the usual 77c in P95 as the CPU is smart enough to scale back the clocks when faced with that kind of load. For being stuck in a 12L case that goes everywhere with a focus on reliability, it's perfect.

Tom's has a new nonsensical article "reviewing" 1.0.0.3ABBA on a 3700X and Taichi, written by a staff member too dense to understand, 2 months after release, that 1.4v is not being put through the CPU while it's under any semblance of actual load. All with the clickbait title implying that 3700X can't hit boost clocks at -180c, when in all likelihood, PEBCAK

Y'all think I should try a 32GB Trident Z kit at the same 3200/C16 to see if the Aorus can do it? I only stepped down to 16GB because the LPX 32GB kit wouldn't work.

Splave is not a staff writer, he's a "professional" overclocker.
That said, he clearly did something wrong, as my 3800X happily boosts to 4,600MHz using an AIO. Obviously only on a couple of cores, but all cores can go beyond 4,500MHz.
I think wonting articles like that, without even trying different boards from at least a couple of manufacturers is also fishy.
Not all boards work the same, as we've seen here and different manufacturers have had varying success with their UEFI's when it comes to performance.
As such, that article from Tom's should mostly be ignored, imho of course...

 

[TheoneandonlyMrK]

Splave is not a staff writer, he's a "professional" overclocker.
That said, he clearly did something wrong, as my 3800X happily boosts to 4,600MHz using an AIO. Obviously only on a couple of cores, but all cores can go beyond 4,500MHz.
I think wonting articles like that, without even trying different boards from at least a couple of manufacturers is also fishy.
Not all boards work the same, as we've seen here and different manufacturers have had varying success with their UEFI's when it comes to performance.
As such, that article from Tom's should mostly be ignored, imho of course...

I think a lot of people having issues are having difficulty because of not realising all the effective parameters, i have now messed about enough to have worked through some conclusions.
Ryzen master if loaded Will take over the clocks for example.
PBO seams to not work as intended since it pushes volts not core speed which ends up counter productive.
Asus pre configured overclocking was made for earlier chips and still works the same so is ineffective on third gen, comparative to gen two that is.
Any auto tune setting will likely reduce top clock.
Default clock
Default tuning hits highest clocks.
Manually setting a max multi is the only way to get an all core boost that You ask for constantly but at a high heat cost.

 

[Mussels]

Front mounted rad?

Yup. The H55 rear mounted exhaust runs the GPU atm and struggles to keep up when both are fully loaded, so i can either lower wattages or lower wattages and switch the coolers around (65W CPU wont overload a 120mm AIO)

 

[lexluthermiester]

Yup. The H55 rear mounted exhaust runs the GPU atm and struggles to keep up when both are fully loaded, so i can either lower wattages or lower wattages and switch the coolers around (65W CPU wont overload a 120mm AIO)

You've likely already given this thought, but a new case might serve you well. One where you can mount the longer rad up top as to exhaust the heat straight out of the case.

 

[Mussels]

You've likely already given this thought, but a new case might serve you well. One where you can mount the longer rad up top as to exhaust the heat straight out of the case.

This current system is a merger between my old build and my VR rig, moved house and needs changed.
Current plan is lower wattage CPU, better case when i find one i like.

(I'm using a lot of old cooling hardware due to a newfound hatred of my AIO's requiring USB connections, when iCUE had some bad releases it was a frigging nightmare to keep everything working correctly. 65W 3700x with a pretty tower air cooler and the H100 on the 1080 could make me happy)

 

[the54thvoid]

I think a lot of people having issues are having difficulty because of not realising all the effective parameters, i have now messed about enough to have worked through some conclusions.
Ryzen master if loaded Will take over the clocks for example.
PBO seams to not work as intended since it pushes volts not core speed which ends up counter productive.
Asus pre configured overclocking was made for earlier chips and still works the same so is ineffective on third gen, comparative to gen two that is.
Any auto tune setting will likely reduce top clock.
Default clock
Default tuning hits highest clocks.
Manually setting a max multi is the only way to get an all core boost that You ask for constantly but at a high heat cost.

PBO doesn't boost clocks, it's not meant to. It allows a longer sustained clock speed, up to the maximum boost clock (not higher) as long as three different power and heat parameters are in check.

 

[tabascosauz]

Splave is not a staff writer, he's a "professional" overclocker.
That said, he clearly did something wrong, as my 3800X happily boosts to 4,600MHz using an AIO. Obviously only on a couple of cores, but all cores can go beyond 4,500MHz.
I think wonting articles like that, without even trying different boards from at least a couple of manufacturers is also fishy.
Not all boards work the same, as we've seen here and different manufacturers have had varying success with their UEFI's when it comes to performance.
As such, that article from Tom's should mostly be ignored, imho of course...

I think a lot of people having issues are having difficulty because of not realising all the effective parameters, i have now messed about enough to have worked through some conclusions.
Ryzen master if loaded Will take over the clocks for example.
PBO seams to not work as intended since it pushes volts not core speed which ends up counter productive.
Asus pre configured overclocking was made for earlier chips and still works the same so is ineffective on third gen, comparative to gen two that is.
Any auto tune setting will likely reduce top clock.
Default clock
Default tuning hits highest clocks.
Manually setting a max multi is the only way to get an all core boost that You ask for constantly but at a high heat cost.

Well, that explains a lot. There's already so much misinformation that goes around regarding voltages on Ryzen 3000, so this kind of "content" from a site like Tom's doesn't help.

I don't think he's got the entirely wrong idea, just that he took a misguided, simplistic approach and turned it into clickbait. Fixed freq is still the best way to bench and explore the voltage limits of your chip because you can focus on Vcore and LLC at a single volt value that matters. The only issue is that all the limits and safeguards come off on fixed freq and your CPU is free to pump out as much heat and power consumption as it would like at whatever voltage you've set. With boost and Vcore offset, I've been finding that most of the instability shows up in light load and high boost, which is hard to get around because conventional stress tests pass with flying colours.

Swede is right, despite being the same version of AGESA there's so much variation in UEFI implementation between vendors, so just because ABBA behaves properly on one board is no indication even another board from the same vendor will perform the same. Add to that the fact that a lot of UEFIs copy-pasted functions like CPB over from Ryzen 2000, not knowing that they would have completely different and possibly unwanted effects (CPB can disable boost entirely) on Ryzen 3000.

 

[Mussels]

I mean, ABBA performs differently on my board depending if my settings are on 'auto' or 'default' because to ASUS, those are different things.
AMD need to force them to be more consistent in the future.

 

[GoldenX]

I mean, ABBA performs differently on my board depending if my settings are on 'auto' or 'default' because to ASUS, those are different things.
AMD need to force them to be more consistent in the future.

I have that same problem with Asus since 2006 on 775.

 

[Xzibit]

I mean, ABBA performs differently on my board depending if my settings are on 'auto' or 'default' because to ASUS, those are different things.
AMD need to force them to be more consistent in the future.

It be nice but I doubt it. Like always boards vendors look to up one another by implementing their own "stock" settings.

 

[TheLostSwede]

I mean, ABBA performs differently on my board depending if my settings are on 'auto' or 'default' because to ASUS, those are different things.
AMD need to force them to be more consistent in the future.

That's really bizarre. And auto is one of those settings that seems to work on voodoo and black magic with a sprinkle of Taiwanese Guai Guai, just to confuse everyone.

Hangry computer - Scandinavia and the World

Webcomic: In Taiwan there's a belief among a lot of people that electronics will only work if you place Kuai Kuai snacks near them. The

satwcomic.com

 

[Mussels]

If/when i get a 3700x, part of me wants to run my big AIO for the GPU and stay with the wraith prism cause it looks pretty

 

[GoldenX]

If/when i get a 3700x, part of me wants to run my big AIO for the GPU and stay with the wraith prism cause it looks pretty

Cooler/Marketing team: "job done".
It sure looks nice.

 

[Mussels]

I always come up with upgrade plans, even if i rarely follow them

Currently pondering 3700x with wraith prism (noise dependant) or a dark rock slim with my ML120 fans
180W TDP design on a 65W CPU should be beyond overkill, and i get to keep my RGB fans

 

[Xzibit]

Incase some are unaware. New Chipset drivers are out.

v1.9.27.1033

Supports:
Windows® 7/10 for
2nd Gen AMD Ryzen™ Desktop Processor
7th-Gen AMD A-Series Processors
AMD Ryzen™ Desktop Processor
AMD X470 Chipset
AMD X370 Chipset
AMD B450 Chipset
AMD B350 Chipset
AMD A320 Chipset
Windows® 10 only for
3rd Gen AMD Ryzen™ Desktop Processor
2nd Gen AMD Ryzen™ Threadripper™ Processor
AMD Ryzen™ Threadripper™ Processor
AMD Ryzen™ Desktop Processor with Radeon™ Vega Graphics
AMD Ryzen™ Mobile Processor with Radeon™ Vega Graphics
AMD X399 Chipset
AMD X570 Chipset
This is an all-new chipset software installer with updated UI, optimized package size, package versioning unique to the chipset drivers, and improved installer performance.
Package Includes:
AMD Chipset Drivers
AMD Ryzen™ Power Plans (required for UEFI CPPC2 in Windows® 10 May 2019 Update)
Contains desktop idle optimizations for 3rd Gen AMD Ryzen™ Processors

 

[robot zombie]

I'm not sure you should go into Ryzen 3000 with the intention of reducing temperatures, as you may find yourself in for an unpleasant surprise. Now, with performance and efficiency in mind, you most definitely should

So far, I really like what 1.0.0.3ABBA has done. Idle is stable on temperatures, and a firmware change now allows HWInfo to read CCD temp (the actual cores portion as opposed to the usual Tdie. Benchmarks aren't being favoured as the multipliers quickly back down from 43-44x to 40x, but I suspect that enabling PBO again would quickly solve that issue for benchmark warriors. Light benches like CPU-Z perform well. On the other hand, 44x is being achieved on roughly half of the cores, whereas ABB would top out at 43.5x on just the fastest core.

End result is that gaming stays quiet around 55-60c with max boost on the cores that need it, while benchmarking is reasonable at 60-70 up to the usual 77c in P95 as the CPU is smart enough to scale back the clocks when faced with that kind of load. For being stuck in a 12L case that goes everywhere with a focus on reliability, it's perfect.

Tom's has a new nonsensical article "reviewing" 1.0.0.3ABBA on a 3700X and Taichi, written by a staff member too dense to understand, 2 months after release, that 1.4v is not being put through the CPU while it's under any semblance of actual load. All with the clickbait title implying that 3700X can't hit boost clocks at -180c, when in all likelihood, PEBCAK

Y'all think I should try a 32GB Trident Z kit at the same 3200/C16 to see if the Aorus can do it? I only stepped down to 16GB because the LPX 32GB kit wouldn't work.

Good stuff... that's what my board's BIOS is packing now. Very reassuring! I would be really curious to see how PBO works out, though.

I say if you've got the kit, run that shit! I dunno if I'd be going out of my way to buy some nice TridentZ, though. But then I'm the guy who wants to put a 3900x in a midrange X370, so... ...I mean it's valuable information

And yeah... Tom's isn't exactly the pinnacle of good information or intrigue these days. Shame because in the past I've really liked their in-depth review style for components - they really cover a lot and get into the reasons for this and that... they were always good at sticking with the times when maybe all of the info wasn't quite converging in all of the different communities. But their editorials are often complete garbage... riddled with the same flavor of misinformation and sprinklings of personality dysfunction common in gaming journalism. Toss it up, keep it moving...

I always come up with upgrade plans, even if i rarely follow them

Same... I had originally meant to do an all new Ryzen 3000 build with liquid and all. Now I'm looking at just keeping most of my existing stuff, motherboard and all, dropping it in a new case (something reasonably liquid-cooling-friendly in case I ever summon my willpower - and not as beat-up as my S340 Elite,) and installing a Ryzen 3000 into that. I have an extra ML140 for a triple-fan case as well as the Dark Rock 4 200w TDP cooler, along with a Corsair Rm650x, so I'm set for any chip that'll run on that board with pretty much any GPU I'd want to run (assuming AMD doesn't come up with a power-sucker I've just gotta have, not likely...) fortunately the stuff I do invest in doesn't always need to be upgraded, so if for whatever reason I can't have it all, I can have at least the main thing I want and not suffer just by living with the rest. AMD is helping big-time with that, letting my old reliable mobo take their newest chips. I had always hoped that would work out back when Ryzen first showed up... got my wish and I'm just trying to be glad I can't fully justify an X570 just yet. Such is life when you have other hobbies and limited income. The moment I find a new girlfriend it's basically all over for me... though maybe this time around I try to find one that actually wants to have a job and make money to help make things happen...

 

[tabascosauz]

Incase some are unaware. New Chipset drivers are out.

v1.9.27.1033

Supports:
Windows® 7/10 for
2nd Gen AMD Ryzen™ Desktop Processor
7th-Gen AMD A-Series Processors
AMD Ryzen™ Desktop Processor
AMD X470 Chipset
AMD X370 Chipset
AMD B450 Chipset
AMD B350 Chipset
AMD A320 Chipset
Windows® 10 only for
3rd Gen AMD Ryzen™ Desktop Processor
2nd Gen AMD Ryzen™ Threadripper™ Processor
AMD Ryzen™ Threadripper™ Processor
AMD Ryzen™ Desktop Processor with Radeon™ Vega Graphics
AMD Ryzen™ Mobile Processor with Radeon™ Vega Graphics
AMD X399 Chipset
AMD X570 Chipset
This is an all-new chipset software installer with updated UI, optimized package size, package versioning unique to the chipset drivers, and improved installer performance.
Package Includes:
AMD Chipset Drivers
AMD Ryzen™ Power Plans (required for UEFI CPPC2 in Windows® 10 May 2019 Update)
Contains desktop idle optimizations for 3rd Gen AMD Ryzen™ Processors

v1.9.27.1033 chipset drivers basically undid most of my improvements resulting from 1.0.0.3ABBA and returned my rig to 1.0.0.3ABB performance, despite being on ABBA. I'm getting sick and tired of AMD's changelog team slapping "idle improvements" on every AGESA and chipset update and telling us nothing else about what was changed, maybe they need a good flogging to remind them that they're on the job.

Benches better on single and multi in CPU-Z (I don't really do much benchmarking as I look for better voltage and temp performance), but so did 1.0.0.3ABB. Obviously, at the cost of idle thermals, which pretty much are indistinguishable from ABB with -0.075V offset.



Also, while practical performance is improved, the CPU multipliers are out of wack again. Only one core is hitting 44x now, as opposed to 3 of them on the old drivers.



The new drivers bring back some of the "snappiness" of ABB when opening programs, kind of a hallmark of Ryzen 3000 over Intel platforms, at this point. I don't mind the performance boost; 1.0.0.3ABBA took away the responsiveness in favour of gaining a calmer idle and the new "idle filter" to stop programs from requesting boost all the damn time.

But the new chipset drivers advertise "desktop idle optimizations" and not "performance improvements", and it appears that they've just done the exact opposite and rolled back most of ABBA's changes.

Ryzen Balanced and Windows Balanced seem to do the same things now.

 

[Roddey]

I backed up before I installed the new chipset drivers. The thing is, that using Macron Reflect it only took 1 minute 19 seconds to restore. I think thats amazing, so fast that its done by the time I can get a glass of water. New nvme drives.

 

[TheLostSwede]

New Gigabyte beta UEFI's are out for some of their X570 boards, F7c in case of the Aorus Master.

GIGABYTE Latest Beta BIOS

Warning Some of beta BIOSes are still undergoing compatibility testing. GIGABYTE is sharing these BIOSes for testing purposes only and are not meant f

forums.tweaktown.com

 

[GoldenX]

If, for any reason, your vendor didn't give you the ABBA goods:

AGESA FW stack patched bioses for 3rd gen

www.overclock.net

 

[GoldenX]

Nvm the prefious post. New official beta bios from MSI with ABBA.

 

[AlienIsGOD]

gave my 2700 a lil rub

3.6ghz on all cores

 

[mstenholm]

gave my 2700 a lil rub

View attachment 133854

3.6ghz on all cores

First off - congtratulation with the huge OC

My 2700Xs was off (power cut) for a longer period. I have a Kill-a-Watt permanent hooked up on one off them and at re-start I noticed that the power draw had gone up from 200 to 260 W. It turns out that that rig decided to run one core at 4200 MHz and the rest on base so instead of 8 times 3850 MHz I got 1 times 4200 and the rest at 3600 MHz at the cost of 60 W. 1x4200 plus 7x3600 => 3675 MHz average. A poor deal. OC'ing the old ones isn't really worth it, at least not for us running them 24/7 @ 100% CPU. Do you have any power consumption numbers?