Does anybody know, what VRM components are used on the ASUS WS Z390 Pro (https://www.asus.com/us/Commercial-S...WS-Z390-PRO/)? I couldn't find anything on google and ASUS support isn't helping out (they say it's board level confidential *lol*). I couldn't find any pictures without the heatsinks or of the backside, so I can't figure out the VRM on my own.
I'm interested in buying a ASUS WS Z390 Pro. But with all that Maximus XI twin-8 stuff going on, i'm a little hesistant on going with ASUS this time.
I would appreciate if you could share any thoughts about that board. I'm looking for a Z390 board with PLX chip, so there is only the supermicro and this one, but supermicro is nearly impossible to get in my country. Are there any pros / cons you would care to point out about the WS Z390 Pro?
Mate I can not answer that as I do not know. What I will say is that the discussions going on about the so called VRM issues are so blown out of proportion as to be funny at times. The Maximus boards have been used to set the most records on HWBOT, pushing the power delivery beyond anything a normal user who OC's their system will ever use without any problems at all.
I have the Extreme board and mates use the Gene and Hero to bench with. Not a single issue with getting the max out of our systems using the power delivery (VRM) as it stands.
What some label a Marketing weenie used I am not going to discuss as it is pointless at times making sense of marketing speak and claims, but on the motherboard specification pages I have not found reference to the VRM claims which caused the debate. They key thing here is that I personally am telling you that you will not have any issue with the power delivery of the current crop of ASUS motherboards and that you can buy the Pro with confidence.
To help us help you - please provide as much information about your system and the problem as possible.
I'm planning to use a pre-binned 9900K because my use case needs very high single / low thread performance as well as good multithread performance. Thats why I will have to OC it as well as possible (within reasonable scale), so VRM performance indeed matters with the high 9900k power demands.
I have a very unusual use case, where one requirement resulsts in wanting to drive 3 seperate M.2 NVMEs as dedicated devices (no RAID) with very high IO counts. I would definitely prefer to put the SSDs on direct CPU attached lanes. Thats what I need the PLX chip for (9900k supports only up to 8x/4x/4x - not enough for 3 SSDs + GPU. And only very few boards support pcie bifurcation, anyway).
Only other possibility would be an AOC with pcie x8 to 4 port pciex4 NMVe M.2 with plx chip (like https://www.supermicro.com/products/...-SHG3-4M2P.cfm). The downside of that solution would obviously be the bottleneck on the x8 pcie side. Thats why I'm considering the WS Z390 Pro, but I can't get any details about the VRM and would like to know a little more about it. I concur that the whole VRM discussion is probably a little over the top, but as it seems to be at least a little bit valid, I would like to clarify this in advance. For example, I read or watched somewhere in a mainboard comparison / review (don't remeber exactly where, would take a little time to find it again) that one of the Maximus XI Hero needed a significantly higher voltage level (of course considering different vdroops) than other boards they compared it with (e.g. Gigabyte Aorus Master) to maintain the same stable clockspeed on the same processor. That of course drove the temps even higher, VRM as well as CPU. They speculated the instability at the same (lower) voltage than on the other boards might be caused by the higher output ripple of the ASUS VRM.
Carlyle2020, thanks, I already watched the review of hardware unboxed. Unfortunately, they only reviewed the Hero and another low end ASUS board, but not the WS Z390 Pro. And as the WS Z390 Pro is outside of the Maximus line and also the marketing speak difffers from Maximus (e.g. they never mentioned the term "twin-8", it might be a different setup.
As for the other question, please see my post #4 above (probably our posts overlapped). Its about direct CPU attached PCIE NMVEs.
Thanks. Well a PLX chip uses what lanes are available and can play the middleman allocating more or less between connected pcie devices. It does not multiply lanes. M.2 SSDs are finally one of the first hardare items almost maxing out 4xpcie lanes. So a 4x connection with PLX toggeling between 2 M.2 SSDs wil not do the trick. So a PLX chip would not get you what you desire. X299, which has more lanes or straight up AMD Boards would help archieve the connectivity that you mention above in theory. But 3x M.2 SSDs on CPU? I think i saw someone wanting to connect 4 x 4x M.SSDs and did not find a solution without compromise. If you are willing to run the GPU with 8x there are more options since PLX can be on a pcie extension board allocating 8x to 2x4x. Linus tried this with his 1 PC with 8x GPUs for 8 Editing Stations Video. But the hardware he used was expensive and not easy to configure. Up until today he failed to get the thing running proper.
Well, in case of the WS Z390 Pro, it multiplexes all 16 CPU driven PCIe lanes between the 4 slots. As I use my graphics card mostly for a triple 4k monitor setup and not gaming or other GPU intensive tasks, the GPU does not use a lot of bandwidth. So almost all of the bandwidth can be used for the 3 NVMEs, more than enough for three PCIEx4-M.2-NMVEs.
In case of an add-on-card, I would use a PCIe slot, that uses x8 PCIe (cutting the graphics card from x16 to x8). So it would be multiplexing the x8 PCIe lanes to three M.2x4-NVMEs. Not ideal, but better than using a passive PCIex8 card with 2 NMVEs (which would need a board with PCIe port bifurcation support to x4+x4 on the second CPU driven PCIe slot, e.g. Hero or Gigabyte Designare) and having to put the 3rd NVME on the PCH lanes. With the Taichi it is quite similar, just that it splits the 16 PCIe lanes to 3 physical slots with x8 (GPU), x4 and x4 instead of one slot with x8 and one other slot with x4+x4 (bifurcated).