B450 MSI Gaming Pro Carbon AC + R5 2600X

[AlwaysHope]

Ryzen + generation interesting to fiddle with, in combo with new B450 platform. After 5+ yrs on AM3+ platform for my daily gamer rig, it's......refreshing! to say the least...lol..

System is as listed in my profile system specs.... although using 2400Mhz Kingston HyperX 2x4GB kit, that I will upgrade in time but want to see how high these ram sticks can OC after getting use to 2600X & it's behavior.

MSI board has only 1st release bios, so within the confines of that, here is some early testing, most on auto settings.

CPU cooler is Coolermaster MA620P with 120mm push fan - Cryorig QF120 Performance model & 120mm pull fan - Fractal Venturi HF-12.
Reason for this unusual fan combo is that is all I had on hand & wanted to improve on the standard CM fans. Both of those fans mentioned are capable of 83.5 cfm.

Ryzen Master Utility v1.4... can only find up to 300 seconds for stress testing with this suite, CPU @ default but RAM @ 2666MHz - 1.26 vdimm as set in bios. Memtest86+ stable

This software suite is only compatible with win 10.

Still exploring this new software package but reminds me a bit of AMD overdrive suite for AM3+ platform.

20 mins of p95 (latest edition) in blend mode... I noticed as soon as the run swings into small ffts, the Vcore temps skyrocket up to 83C + so i shut it down as that doesn't seem kinda healthy. More research needed if this app is reliable for Zen+ stress testing today.

More results to come obviously, only early days so far with this hardware setup...

 

[biffzinker]

Seems to be operating the same as mine with regard to XFR2 & Precision Boost Overdrive (PBO). Tried stress testing in IBT but stopped it when temps were similar to your P95 result.

Edit: Took it for a spin in any games or messed with anything now that you have the extra cores/threads (SMT?)

Edit 2: Looks like your 2600X requires less voltage compared to mine between 4.1-4.250 GHz. Might be able to do better OCing towards the higher boost frequencies.

 

[AlwaysHope]

Seems to be operating the same as mine with regard to XFR2 & Precision Boost Overdrive (PBO). Tried stress testing in IBT but stopped it when temps were similar to your P95 result.

Edit: Took it for a spin in any games or messed with anything now that you have the extra cores/threads (SMT?)

Edit 2: Looks like your 2600X requires less voltage compared to mine between 4.1-4.250 GHz. Might be able to do better OCing towards the higher boost frequencies.

Yeah, seems IBT is better & quicker way to determine full load stability atm. Not going to fiddle with CPU OC until I get the ram OC stable. Enabled XFR2 in bios, but that's the only thing changed from defaults.

I think these Kingston sticks will do up to 3200Mhz, That's what I've seen other users of these sticks do with Ryzen systems. But for now 2800Mhz @ 16-17-17 1.30v testing atm.

Loaded up my 1 of my fav games - FO4, last night with HWiNFO64 v5.86-3480 & noticed that game engine using 9 threads. Temps between 50-60C.

This MSI bios needs fine tuning though, can't disable WiFi or bluetooth at this level.

 

[AlwaysHope]

Ok, got this Kingston HyperX 2400Mhz 8GB kit running stable & by that I mean stable for my needs @ 2933Mhz now with 1.35Vdimm. Tried to keep the main timings reasonably tight.
Well aware how DDR4 bandwidth benefits Zen+ architecture but not all of us have a money tree in the garden. So this is where OC experimenting comes in REAL handy!

Something I have to mention is how much smoother playing my currently fav game is - FO4, seriously without bragging, the difference is like between night & day compared to my old AM3+ platform. That older rig was built in mind with the devs of that games recommended specs for AMD systems too. That is with a FX-9xxx series CPU back in 2015 although I had OC FX-8350 at those levels. That games engine loves cpu power & lots of threads.

Anyway, latest results so far...

PCMark10, Imo good for testing I/O of many aspects in the system. Beat 4K system with 2015 R9 Nano too...lol... & no M.2 storage.

IBT on very high, 10 runs, CPU stock, RAM@2933Mhz

Ryzen Master Utility stress test with Kingston HyperX 2400Mhz ram OC to 2933Mhz..

Cinebench R15.038, all tests.

Still much much more to learn with this platform...

 

[ne6togadno]

https://www.techpowerup.com/download/ryzen-dram-calculator/

 

[AlwaysHope]

https://www.techpowerup.com/download/ryzen-dram-calculator/

Well aware of that app, but not working for me. I mean how do you know what values to enter for the sub timings if they set to auto in bios? Datasheet from Kingston for my particular ram does not have all that info. > https://www.kingston.com/dataSheets/HX424C15FB_4.pdf
Considering how cheap I got this ram, I'm impressed with 533MHz OC in bandwidth. Can't complain really.

Boots at 16-17-17-40@ 3000Mhz, memtest86+ & Memtest64 passes all good, but fail IBT.

Tried entering CL17 for 3000MHz ram testing, but other column of bios reports CL18 entered.... go figure.. booting into windows, HWiNFO & CPUZ report CL18. Seems v1.00 bios does not like or want CL17 with these sticks for 3000MHz.

So, booting with 3000MHz 18-17-17-40 & running Aida64 + PCMark10 reports negligible benefits.

For time being, 2933MHz @ 16-17-17-40 is where it's at for these ram sticks.

Current usage pattern for this platform does not benefit from manual OC with CPU. XFR2 is good enough for my daily gaming needs atm.

 

[ne6togadno]

you need thaiphoon burner to get info for your ram.
some screenshots and tutorial like review for the calculator https://www.techpowerup.com/forums/threads/ryzen-dram-calculator-1-3-0.246327/

 

[Xzibit]

So whats the background picture with the lace, How many threads ?

Nice build, Ryzen Master looks interesting. Can you run profiles per app. Possible to underclock or disable cores ?

 

[AlwaysHope]

you need thaiphoon burner to get info for your ram.
some screenshots and tutorial like review for the calculator https://www.techpowerup.com/forums/threads/ryzen-dram-calculator-1-3-0.246327/

Nah, use The Stilt's Ryzen Timing Checker, all's good.

Thaiphoon Burner is shareware, I'm allergic to that type of software. Not handing over $ for paid version.

So whats the background picture with the lace, How many threads ?

Nice build, Ryzen Master looks interesting. Can you run profiles per app. Possible to underclock or disable cores ?

2600X has 12 threads, as indicated in HWiNFO64 & official AMD specs.

There are profiles you can fiddle with in RMU, haven't bothered much yet, Options in bios to disable cores.

Time to update bios to 2nd release.

One thing I have to comment on is the onboard sound & it's Nahimic app, the option for loudness control whilst gaming is pathetic & weak. My old Sabertooth board had better control with this aspect.

 

[AlwaysHope]

Ok, latest 2nd release bios does not approve of loading saved OC profiles from earlier version...
They have changed precision boost overdrive options as seen in this screenshot.

 

[biffzinker]

If you update the firmware to the latest version you lose your prior profile settings. I've had it happen with Gigabyte, Microstar International, and ASUStek.

 

[AlwaysHope]

If you update the firmware to the latest version you lose your prior profile settings. I've had it happen with Gigabyte, Microstar International, and ASUStek.

Yeah, its annoying... didn't happen on my Asus Sabertooth 990FX R2.0 board last time I upgraded the bios.
Reading up on the RAMDISK option sounds interesting. I like that concept but will have to upgrade ram capacity no doubt.

 

[AlwaysHope]

Manual OC attempt @ 4.4GHz in Cinebench - all tests, of course after RMU test for stability.

 

[AlwaysHope]

XFR2 left on auto in bios is over volting something stupid.

Here's what HWiNFO64 v5.86-3480 reports with just 10 mins of game play in FO4.

 

[ne6togadno]

 

[AlwaysHope]

A

Thx, already aware of that, some of the options in that bios are not available on this B450 board, but generally identical bioses.

I'll get around to that when time available, but for now still fine tuning RAM.

 

[robot zombie]

XFR2 left on auto in bios is over volting something stupid.

Here's what HWiNFO64 v5.86-3480 reports with just 10 mins of game play in FO4.

View attachment 106160

Woah! Seems a little high.

But as I'm understanding it both PBO and XFR have a max allowance. It'll pump each core's clock as high as it can within power/temp specs according to sensor data. So if it's reading low enough temperatures, it'll continue to bump frequency and voltage right up to the max allowance. Looks like the max it'll allow is 4.3mhz @ 1.5v. I don't know the exact numbers, but I think that's reasonable, just looking at what most people tend to manage at max. It's supposed to be hard coded. Ryzen 2 will not auto-overclock past a certain point. May even vary from chip to chip, depending on how it tests at the factory.

I wonder how transient those maxes are, spread out over time, and across different cores. I'm almost betting they're little bursts on individual cores. And those little bumps probably lend the individual cores stability in those moments when a manual OC at a lower overall voltage might crap out.

You're right though, it seems wonky to me too. But then I can tell you in my experience manually OCing my straight 2600 that those voltages don't look out of line for what's needed at those clocks. So it may actually be working as intended. Just so happens that Ryzen 2 caps out at 1.5 on things like XFR and PBO. Don't quote me on that. I just know that they advise not to go over 1.5v, so it makes sense that auto overclocking would also cap out there. I'm also gonna assume that it probably limits current so that it isn't possible for all cores to hit the voltage limit at once, or maintain it for too long... ...not that they could because temperatures would surge and it'd throttle all on its own if you were shooting through 1.5v in any remotely continuous way. Depends on your cooling situation, I suppose. But I think for most setups it shouldn't be possible for that to occur.

Just throwing my thoughts out there. I'm only beginning to play with my 2600 myself. For all I know I'm being too trusting. But I also think we are getting to a point where auto-overclocking is actually getting pretty smart. One side of me thinks "eh, seems okay" while the other part of me doesn't like the idea of a machine trying to do what for me would amount to a very careful judgement call. But hey, ideally they should be able to do it better than us! Running a manual OC at 1.5v and swinging to that on a "smart" OC aren't quite the same.

 

[AlwaysHope]

Woah! Seems a little high.

But as I'm understanding it both PBO and XFR have a max allowance. It'll pump each core's clock as high as it can within power/temp specs according to sensor data. So if it's reading low enough temperatures, it'll continue to bump frequency and voltage right up to the max allowance. Looks like the max it'll allow is 4.3mhz @ 1.5v. I don't know the exact numbers, but I think that's reasonable, just looking at what most people tend to manage at max. It's supposed to be hard coded. Ryzen 2 will not auto-overclock past a certain point. May even vary from chip to chip, depending on how it tests at the factory.

I wonder how transient those maxes are, spread out over time, and across different cores. I'm almost betting they're little bursts on individual cores. And those little bumps probably lend the individual cores stability in those moments when a manual OC at a lower overall voltage might crap out.

You're right though, it seems wonky to me too. But then I can tell you in my experience manually OCing my straight 2600 that those voltages don't look out of line for what's needed at those clocks. So it may actually be working as intended. Just so happens that Ryzen 2 caps out at 1.5 on things like XFR and PBO. Don't quote me on that. I just know that they advise not to go over 1.5v, so it makes sense that auto overclocking would also cap out there. I'm also gonna assume that it probably limits current so that it isn't possible for all cores to hit the voltage limit at once, or maintain it for too long... ...not that they could because temperatures would surge and it'd throttle all on its own if you were shooting through 1.5v in any remotely continuous way. Depends on your cooling situation, I suppose. But I think for most setups it shouldn't be possible for that to occur.

Just throwing my thoughts out there. I'm only beginning to play with my 2600 myself. For all I know I'm being too trusting. But I also think we are getting to a point where auto-overclocking is actually getting pretty smart. One side of me thinks "eh, seems okay" while the other part of me doesn't like the idea of a machine trying to do what for me would amount to a very careful judgement call. But hey, ideally they should be able to do it better than us! Running a manual OC at 1.5v and swinging to that on a "smart" OC aren't quite the same.

Yeah, Buildzoid when reviewing the bios of MSI X470 Gaming Pro Carbon clearly indicated he thought the same about how dumb the A.I is with XFR2. Gotta agree with his commentary on how it just over volts ridiculously high.
If you observe the data from my post on Cinebench @ 4.40GHz in this thread & look at the Vcore there, it proves this point. Even checking for stability with RMU at those settings emphasized this.

 

[robot zombie]

Yeah, Buildzoid when reviewing the bios of MSI X470 Gaming Pro Carbon clearly indicated he thought the same about how dumb the A.I is with XFR2. Gotta agree with his commentary on how it just over volts ridiculously high.
If you observe the data from my post on Cinebench @ 4.40GHz in this thread & look at the Vcore there, it proves this point. Even checking for stability with RMU at those settings emphasized this.

Ahh... didn't put that one together. Well I guess bob's my uncle. Looks like there are some kinks to work out there. Cheers on those results, btw. Very nice.

I will say personally I still am not a fan of the auto stuff in general. Not just because I like to tweak things but just because it occasionally sets weird parameters.

Be nice if they could work in some sort of simple peak voltage offset. Basically just have the whole thing run like it would otherwise, only pretending the peak voltage is higher than it actually is. Let it do its thing and tweak the offset until you bottom out on stability under max loads. Idk, easier than rebalancing everything in time. How many BIOS updates would it take to get it actually working right?


I feel like stuff like this sounds really good on paper. And honestly, it should actually be doable to look at a chip's architecture, node size, etc and reliably predict what combinations of parameters will work, and then select the most optimal power/temp/performance balance. Even better, take the factory testing and maybe some more focused testing, and intersperse it with the theoretical stuff and you could eventually have a very robust algorithm. We have the means to accomplish things like that. Implementation just needs work, I think. Sensors might be a limitation, too. And of course there's the question of can it be compiled quickly enough to still be relevant for each new CPU that comes out.

 

[AlwaysHope]

Ahh... didn't put that one together. Well I guess bob's my uncle. Looks like there are some kinks to work out there. Cheers on those results, btw. Very nice.

I will say personally I still am not a fan of the auto stuff in general. Not just because I like to tweak things but just because it occasionally sets weird parameters.

Be nice if they could work in some sort of simple peak voltage offset. Basically just have the whole thing run like it would otherwise, only pretending the peak voltage is higher than it actually is. Let it do its thing and tweak the offset until you bottom out on stability under max loads.

Agree with you about how wonky auto OC is & was in the past. With newer Zen+, its still pretty much the same imo.
Perhaps future bios updates will install cpu voltage offset, but this board & from what I see with X470 version of this series with MSI still don't have that option.
Still getting use to this 2nd release bios though.

 

[AlwaysHope]

Update on imo, poor sound quality with the onboard device.
Had to get rid of Realtek & Nahmic AIO drivers from MSI, just using Win 10 driver & the sound quality is substantially better + I get options to control aspects like loundness & bass management better imo. This is all via my Razer Tiamat 7.1 headphones.

So glad of that, I was despising investing in dedicated sound card...

 

[AlwaysHope]

Bios screenshots show limitaions of options as opposed to what one would get with X470 version of this board. In stating that though, there is minimal difference at best. Perhaps future bios updates will bring new options? but for now this is with 2nd official bios. These are my settings for attempts with manual OC.

1.55v peak cpu voltage, I will endeavour to keep this as low as possible especially with tower push/pull air cooling in enclosed case.

2.00v max Vdimm opition. Honestly... no way I'd like more than 1.35v for daily use.

 

[biffzinker]

My MSI B350 board has a upper limit of 1.4 for manual CPU voltage adjustment. Everything on Auto for the CPU with XFR and now PBO the voltage climbs up to 1.5 mV.

 

[Dinnercore]

Those 1.5V volts are usually just small peaks tho and totally ok. It does that on stock settings too, just watch your temps. If it would really apply straight 1.5 for more then a splitsecond on your cores, your temps would reach throttle point. I had my 1800x running at manual 1.4875V with highest LLC = 1.502V and it hit around 80°C under a 280mm AiO.

On DDR4 you don´t get as hot as e.g. DDR2 FD-DIMMs and they can take some volts. I use 1.4 - 1.45V daily and the 2V wouldn´t kill it immediatly. Maybe within 100 hours or so

What I don´t like is that max Vcore setting. Why 1.55V? That is enough to kill it within 1-2 weeks when manually overclocking and being stupid, so it does not really protect the end user. And it is to low to be useful for any hardcore OC. Which you wouldn´t really do on this board in the first place, ok. But still it´s an odd point to be at for me.

 

[AlwaysHope]

Those 1.5V volts are usually just small peaks tho and totally ok. It does that on stock settings too, just watch your temps. If it would really apply straight 1.5 for more then a splitsecond on your cores, your temps would reach throttle point. I had my 1800x running at manual 1.4875V with highest LLC = 1.502V and it hit around 80°C under a 280mm AiO.

On DDR4 you don´t get as hot as e.g. DDR2 FD-DIMMs and they can take some volts. I use 1.4 - 1.45V daily and the 2V wouldn´t kill it immediatly. Maybe within 100 hours or so

What I don´t like is that max Vcore setting. Why 1.55V? That is enough to kill it within 1-2 weeks when manually overclocking and being stupid, so it does not really protect the end user. And it is to low to be useful for any hardcore OC. Which you wouldn´t really do on this board in the first place, ok. But still it´s an odd point to be at for me.

Yeah, I get that it's small peaks but unnecessary imo, however I'm reasoning that AMD allow that because of variations in consumer grade silicon quality for their SKUs, afterall its better for them to cover their assess then if something doesn't work as marketed.

Well that's the thing with DDR4 & even back when DDR3 was the current flavor too. I mean those heatsinks they include with most desktop ram is absurd from a technical point of view because those generations of ram don't get as hot as DDR2 or even DDR did back in their days. Unless one is trying to establish world records or something... still, its the appearance of the product & clever marketing to help flog them as to why they include heatsinks for DDR4 ram when its Vdimm by default is very low compared to previous gens of RAM. They must think one is going to cram them into SFF boxes & place the box next to a heater or something..... lol... seriously though, its quite absurd when I see RAM companies flogging ram and glorifying the heatsinks... on DDR4 ram. Even on older DDR3 it was getting long in the tooth that marketing strategy.

I have no idea why MSI have set that DDR volt limit, but keep in mind its only 2nd release bios so far. It's a gamers board, not for LN2 world record achievements... lol... one would use uber high end X470 for non TR series anyway if they were to go down that route.

As for the Vcore limit... I have no motivation whatsoever even getting close to that for my usage. 1.50 would be my limit & then that's for benchmarking. But I can see the temps climbing something fierce with tower air cooling.

Anyway, as I mentioned above in previous post for daily use, 1.35v should be the limit... with emphasis on "daily use", that is NOT setting benchmark records.

My MSI B350 board has a upper limit of 1.4 for manual CPU voltage adjustment. Everything on Auto for the CPU with XFR and now PBO the voltage climbs up to 1.5 mV.

I'd say thats probably because of VRM limitiations? not sure why it peaks at 1.5mv though...