Sapphire's Vega Nano

[Bananasplit_00]

Just now, Notional said:

A Kaby Lake G part, heck even the new Ryzen APU's are much more powerful than any half height card though. I still don't see the point in such a build. It's so limited in what parts you can use, and the performance with that thermal envelope, it's really nonsensical to me.

since when is the Ryzen APU stronger than a 1050TI(and Kabylake G too, but i dont know the performance of that too wellr)? and you can use a full height GPU in the second alternative i put forward, or do the first one but with a full height card. the point would be to push extremes, i for sure would want one

[TheRandomness]

56 minutes ago, Bananasplit_00 said:

IMO they shouldnt, someone needs to make an ITX case that really pushes more than the CPU heatsink size down

Hahahahaha. Check my signature. Specifically, the first spoiler in the spoiler.

[Bananasplit_00]

22 minutes ago, TheRandomness said:

Hahahahaha. Check my signature. Specifically, the first spoiler in the spoiler.

oh has stuff been hapaning there, guess follows broke again because i havent goten any updates

[bobhays]

2 hours ago, Bananasplit_00 said:

since when is the Ryzen APU stronger than a 1050TI(and Kabylake G too, but i dont know the performance of that too wellr)? and you can use a full height GPU in the second alternative i put forward, or do the first one but with a full height card. the point would be to push extremes, i for sure would want one

kaby lake with Vega GH is faster than 1050ti, the Vega GL is just a bit slower I think.  Like was mentioned before you can get the new Hades Canyon NUC for that high performance small form factor that you're looking for. 

[grayperview]

I thought this was going to be the norm with HBM memory, with the need to have memory chips spread over the PCB, This card in theory should run nice and cool, given the unobstructed fan blowing through the heat sink.

[leadeater]

17 hours ago, TheRandomness said:

I'm just a little confused about that VRM sectioning though, I think they got phase 3 and 4 mixed up, and it should be that phase 4 should be the bottom inductor, leading to a 6+1+1 arrangement around, for core + something + HBM2.

I would have said 7 (doubled) + 1. Looks very similar to every other Vega card (12 (doubled) + 1) just with less phases for the GPU. Can't really tell much about the specifics of the VRM design without a picture of the back and the voltage controller.

[TheRandomness]

4 hours ago, leadeater said:

I would have said 7 (doubled) + 1. Looks very similar to every other Vega card (12 (doubled) + 1) just with less phases for the GPU. Can't really tell much about the specifics of the VRM design without a picture of the back and the voltage controller.

Having an odd amount of base phases is never done. Also, Sapphire has kept symmetry with the other two pairs of core phases, so why do they break it there?

[leadeater]

9 minutes ago, TheRandomness said:

Having an odd amount of base phases is never done. Also, Sapphire has kept symmetry with the other two pairs of core phases, so why do they break it there?

The base is even however the 8th output is not used, it'll be 4 doubled but that's only a guess. Need to see the back.

[TheRandomness]

2 minutes ago, leadeater said:

The base is even however the 8th output is not used, it'll be 4 doubled but that's only a guess. Need to see the back.

I’m going to keep with the 6+1+1 stuff. Maybe it’s actually 2 phase memory?

 

EDIT: Forgot what the PCB looked like

[leadeater]

3 minutes ago, TheRandomness said:

I’m going to keep with the 6+1+1 stuff. Maybe it’s actually 2 phase memory?

 

EDIT: Forgot what the PCB looked like

Seems this is the back, haven't looked at it yet. Doing that now.

 

[leadeater]

@TheRandomness

I can see what looks like 4 doublers but the back doesn't match the front, still confused lol.

[TheRandomness]

Just now, leadeater said:

@TheRandomness

I can see what looks like 4 doublers but the back doesn't match the front, still confused lol.

I think the doubling is just to provide the doubled MOSFET count per phase, maybe? I’m trying to get a picture off of the reddit user without the phase markings for Buildzoid, but they’re not responding

[leadeater]

4 minutes ago, TheRandomness said:

I think the doubling is just to provide the doubled MOSFET count per phase, maybe? I’m trying to get a picture off of the reddit user without the phase markings for Buildzoid, but they’re not responding

This will explain why there are two MOSFETs per phase, basic explanation is they work in tandem.

 

Quote

So now that you've read that lets expand that and apply it to computer VRMs. First of all the fly wheel circuit uses a diode. This is really inefficient and massively limits the maximum current through put so in computer VRMs you will find instead of the diode what is called a low side MOSFET. This MOSFET is only on when the High side MOSFET(the component labeled switching transistor) is off or else you would get a short circuit. This low side MOSFET handles the bulk of the current that flows through your load(CPU/GPU core RAM chip...) so these MOSFETs are the most important when building a powerful VRM. Low side and high side MOSFETs typically have current handling capabilities between 20 and 60A at 125C° case temperature.

 

The article I linked shows the PWM being fed directly to the high side MOSFET. In computer VRMs the MOSFETs used have a rather large gate capacitance. Meaning that if you want to switch them on you need to let them charge up. If you want them to charge quickly you will need to supply a current greater than what PWM controllers can provide. So to supply the current a driver MOSFET is used. This MOSFET is typically capable of only handling currents smaller than 10A and can be switched on and off directly by the PWM signal coming out of the PWM controller. The driver MOSFET is not very important to a VRMs current handling capability but they are a key component of computer VRMs so you should know about them.

http://cxzoid.blogspot.co.nz/2015/04/what-makes-good-vrm.html

 

High side = fast switching driver

Low Side = main current supplier, needs high side to switch on and off fast

 

And the job of doublers

 

Quote

So how many phases does your motherboard/GPU have? No more than 8.

8 is the largest number of phases that any PWM controller currently used in computers can produce. So how do we have VRMs with 10 12 14 16 20 24 and 32 phases? Doublers.
Doublers are are specially ICs that take one PWM signal and split it into 2. In the process they cut switching frequency in half but they do give you more phases so you do get the extra current capability and get lower operating temps but don't gain anything in terms of voltage ripple suppression. Another trick motherboard manufactures use  that I hate is putting stuff in parallel. There is a good way to do it where they put extra MOSFETS in one phase which basically creates a "super phase" if they are using high end MOSFETs but more often than not they double the number of inductors. This means that inexperienced buyers who count inductors to get phase counts can easily be fooled into thinking that a board has 8 phases when in reality it only has 4 but with each phase having 2 inductors. having 2 inductors on 1 phase is completely pointless. It does nothing that a single inductor couldn't do other than looking more complex than it is.

http://cxzoid.blogspot.co.nz/2015/04/what-makes-good-vrm.html

[TheRandomness]

4 minutes ago, leadeater said:

This will explain why there are two MOSFETs per phase, basic explanation is they work in tandem.

 

http://cxzoid.blogspot.co.nz/2015/04/what-makes-good-vrm.html

 

High side = fast switching driver

Low Side = main current supplier, needs high side to switch on and off fast

 

And the job of doublers

 

http://cxzoid.blogspot.co.nz/2015/04/what-makes-good-vrm.html

I meant more of having the phases groups, so two high sides and two low sides per 'phase'. Though, the doublers don't agree with this, but I don't agree with the odd phase count. 

[leadeater]

8 minutes ago, TheRandomness said:

I meant more of having the phases groups, so two high sides and two low sides per 'phase'. Though, the doublers don't agree with this, but I don't agree with the odd phase count. 

Not sure on the phase counts either but the hi/low sides sit after the doubler, if being used. The PWM voltage controller is fully configurable so it can give out odd or even number of outputs, the odd part is that left over single phase since if you are doubling you will always get an even resultant phase count. Only thing I can think of is one of the PWM outputs is not being doubled.

 

Edit:

Quote

A perfect example of all of these is the MPOWER MAX motherboard I bought. Here are photos detailing it's VRM design that looks like a 20(20 low and high side MOSFETs and 20 inductors) is driven like a 10(10 driver MOSFETS) and is only fed by 5 PWM(5 doubler ICs and the PWM controller is an 8 phase IR running in 5 phase mode) signals before the doublers.

 

[TheRandomness]

3 minutes ago, leadeater said:

Not sure on the phase counts either but the hi/low sides sit after the doubler, if being used. The PWM voltage controller is fully configurable so it can give out odd or even number of outputs, the odd part is that left over single phase since if you are doubling you will always get an even resultant phase count. Only thing I can think of is one of the PWM outputs is not being doubled.

 

Edit:

 

I do remember Buildzoid once mentioning that you must either go all or nothing with the phases being doubled. This is annoying. I sent him that annotated picture and he said that he'd do a PCB breakdown if it weren't incorrectly labelled, so yeah.

[leadeater]

3 minutes ago, TheRandomness said:

I do remember Buildzoid once mentioning that you must either go all or nothing with the phases being doubled. This is annoying. I sent him that annotated picture and he said that he'd do a PCB breakdown if it weren't incorrectly labelled, so yeah.

I don't see why you would have to, the PWM controller has no idea if it's being doubled or not. If you choose not to double one however you will increase the current on that phase and imbalance is bad.

 

Seems I'm not the only person wondering about this as well.

 

Quote

jstefanop1 • 41 points • submitted 9 days ago

Some of you wanted a pic of the Vega Nano PCB/VRMs so here it is in all its glory:

• Vega 56 epoxied die, with Samsung HBM2
• Essentially its the exact same setup as the reference board, except its cut down to 7 VRMs instead of 12
• 4 Phases instead of 6 (3 phases doubled, one single ?weird)
• Same IR 35217 controller with IR 3598 driver and IR 6894/6811 VRM combo
• On-Semi 4C85N DualFet for HBM2

Nitro 64 Bios Flash works, 1200 HBM2 clocks and 1600+ core clocks. Seems like the VRM/Large pulse heatsink can handle the power since this is a 56 die.

Almost 24k in Firestrike…not bad for a pint sized GPU

https://www.3dmark.com/3dm/25926316

https://www.reddit.com/user/jstefanop1

[TheRandomness]

15 minutes ago, leadeater said:

I don't see why you would have to, the PWM controller has no idea if it's being doubled or not. If you choose not to double one however you will increase the current on that phase and imbalance is bad.

 

Seems I'm not the only person wondering about this as well.

 

https://www.reddit.com/user/jstefanop1

I can't wait for the card to come out, then, if it does  I'd love to get my hands on one, and figure out exactly what that VRM is. Also, that reddit user does say that his annotations are correct, which I just don't believe.

[leadeater]

25 minutes ago, TheRandomness said:

I can't wait for the card to come out, then, if it does  I'd love to get my hands on one, and figure out exactly what that VRM is. Also, that reddit user does say that his annotations are correct, which I just don't believe.

Hmm what if it's not being doubled at all. If it's using the 8 channel PWM controller it could be 7 straight GPU phases + 1 HBM.

[TheRandomness]

1 minute ago, leadeater said:

Hmm what if it's not being doubled at all. If it's using the 8 channel PWM controller it could be 7 straight GPU phases + 1 HBM.

I mean that'd make sense. Vega does need all the phases it can get