Production FormD T1 Classic (READ FIRST POST)

[bynio]

Hi @fabio, what are the default settings for these on the 3950x? (now I'm showing just how little I know about the newer AMD processors) I will add the default to the spreadsheet when I update it.

You could start with 1usmus power plan, then reduce CPU clock oscillations (I'm not doing that one) and then optimise PBO limits.

What you mean?
For the "Default" AMD 3950X, in the BIOS I have the VCORE set to NORMAL (NOT AUTO), RAM at 3600MHz and, and PBO DISABLED.

To be able to TWEAK the Power Limit, you have to ENABLE the PBO in BIOS and then in Ryzen Master I set the Power Limit.

For the OC, I use the Manual OC in Ryzen Master and Setup the Core Speed and the Voltage there.

For the ECO mode, I just activate it inside the BIOS.

I'm running on Auto, around 1.33V idle and drops to 1.23V on higher power consumption. Never heard of switching to Normal. Can you elaborate? Is it only Core Voltage or other too?

 

[fabio]

You could start with 1usmus power plan, then reduce CPU clock oscillations (I'm not doing that one) and then optimise PBO limits.

I'm running on Auto, around 1.33V idle and drops to 1.23V on higher power consumption. Never heard of switching to Normal. Can you elaborate? Is it only Core Voltage or other too?

I do that only to the core voltage. I've seen that in AUTO IT gives more Voltage to the CPU.

I using the 1usmus Power Plan on Windows! Sorry, I forgot to say it.

 

[gregbiv]

Guys, do you have any good solution to keep VRAM cool? I saw some attempts of getting monoblock but I think there will be lots of mounting issues (also there is no dedicated monoblock for my gigabyte x570) Another fan on duct tape or a glue blowing on top of the watercooled cpu block isn't looking that good. Is there any other options for this case? Like small fans on the bottom pushing air to the top?

 

[bynio]

Guys, do you have any good solution to keep VRAM cool? I saw some attempts of getting monoblock but I think there will be lots of mounting issues (also there is no dedicated monoblock for my gigabyte x570) Another fan on duct tape or a glue blowing on top of the watercooled cpu block isn't looking that good. Is there any other options for this case? Like small fans on the bottom pushing air to the top?

Aorus has decent VRM temps (compared to Strix), Optimum-Tech tested monoblock and it wast overall better, more airflow helps.

 

[fabio]

Never had any issue on the VRM, with the Asus X570 itx or Gigabyte, which are pretty cool!
The Max temp I got, in general, was 70C, WAY BELOW the safe limit.

 

[gregbiv]

thanks, @fabio, good to know! I am still waiting for the final water cooling part to arrive and currently, I am using the case with stock fans
looks like it will take forever for highflow v2 to appear in stock

 

[ellroy80]

What you mean?
For the "Default" AMD 3950X, in the BIOS I have the VCORE set to NORMAL (NOT AUTO), RAM at 3600MHz and, and PBO DISABLED.

To be able to TWEAK the Power Limit, you have to ENABLE the PBO in BIOS and then in Ryzen Master I set the Power Limit.

For the OC, I use the Manual OC in Ryzen Master and Setup the Core Speed and the Voltage there.

For the ECO mode, I just activate it inside the BIOS.

Hi @fabio, I was going to put the "default" settings in the spreadsheet, until I realised it was unnecessary. I've never owned a Ryzen system, and it's a looooong time since I've built an AMD system. A lot of the terminology is new to me. Clearly, I have some research to do!

Anyway, on with the latest spreadsheet. Just to note, I am using one of the default conditional formatting settings in Excel to produce the colour range. For all component temperatures EXCEPT for the CPU, all combinations are included in a single colour range for that component. For the CPU temperatures, the ECO mode, PBO 120W, and PBO 130W tests were given a different colour range from the rest of the CPU temperatures because those three temperatures are so much lower than the rest. Including all of them in a single colour range produced skewed results.


MEASURED TEMPERATURES



DELTA TEMPERATURES (Component Temperature - Ambient Temperature)


Unsurprisingly, the ECO mode produces the lowest CPU temperatures. It also results in the lowest GPU, VRM, and chipset temperatures. However, it doesn't beat the SSD temps of the default CPU, exhaust/exhaust combination. The two PBO alterations aren't increasing component temperatures by much (other than the VRMs), but result in much lower CPU temperatures when compared to the default setting. Clearly, a little tweaking does wonders for system temps. What is the difference in performance?

 

[ellroy80]

I don't trust though in the capability of the L12s without the fan, it is not meant to work in that way!

That's fair enough.

Noctua is working on 1 CPU cooler Passive for 95/120W CPU! If I will teardown the system again I will test it anyway!

Have you seen the size of that passive cooler?! There's no way it'll fit in the T1!! Hahahaha!

 

[JRey]

I'm looking at the teardown of the 3080 and it looks like the backplate will be required since VRMs are found on the back of the PCB.

NVIDIA GeForce RTX 3080 Founders Edition Review - Must-Have for 4K Gamers

NVIDIA's new GeForce RTX 3080 "Ampere" Founders Edition is a truly impressive graphics card. It not only looks fantastic, performance is also better than even the RTX 2080 Ti. In our RTX 3080 Founders Edition review, we're also taking a close look at the new cooler, which runs quietly without...

www.techpowerup.com

With that said, I had issues installing my R9 290x and 295x2 in the T1 (2 slot mode) with the backplate on (EKWB). The backplate on the 3080 FE seems a little thicker too. I'm getting worried that the card won't fit.

Thoughts?

 

[Wahaha360]

2020-09-17: PCIe 3.0 vs PCIe 4.0 on RTX3080 = *NO material difference = more time to work on PCIe 4.0 Riser ?

TECHPOWERUP: NVIDIA GeForce RTX 3080 PCI-Express Scaling

So if you're on a desktop platform with PCIe Gen 3 x16, relax. There's little to no performance lost at any resolution. PCIe generational scaling used to matter when multi-GPU was relevant and there was a likelihood of running a graphics card at x8 bandwidth—not so anymore. You're on a PCIe Gen 4 platform, you get the freedom to run the RTX 3080 at Gen 4 x8 with no performance lost (since it is the same bandwidth as Gen 3 x16), and may drop a sexy NVMe RAID card into the second slot.

 

[TSOF]

Yep - I think if it's going to make a difference, it wont be until Direct Storage comes into play... so probably not for 18 months at least.

No rush.

 

[PKAWA]

I'm looking at the teardown of the 3080 and it looks like the backplate will be required since VRMs are found on the back of the PCB.

NVIDIA GeForce RTX 3080 Founders Edition Review - Must-Have for 4K Gamers

NVIDIA's new GeForce RTX 3080 "Ampere" Founders Edition is a truly impressive graphics card. It not only looks fantastic, performance is also better than even the RTX 2080 Ti. In our RTX 3080 Founders Edition review, we're also taking a close look at the new cooler, which runs quietly without...

www.techpowerup.com

With that said, I had issues installing my R9 290x and 295x2 in the T1 (2 slot mode) with the backplate on (EKWB). The backplate on the 3080 FE seems a little thicker too. I'm getting worried that the card won't fit.

Thoughts?

That is concerning.
If you aren't in a rush to get the card I will try it.

 

[TSOF]

I haven't bought a GPU at launch for a really, really long time... (at least a decade).
Are we likely to see much (if any) performance improvement from driver optimisations in the next month or two?

Can't recall....

I've only just realised that there is a series of LED's on the "x" portion of the card (the bit you'll see through the mesh).

 

[JRey]

That is concerning.
If you aren't in a rush to get the card I will try it.

I'm going to attempt to buy a card tomorrow BUT I won't be opening it. I'll attempt to buy a 2nd card in mid-October (return the launch day card) to take advantage of the 30-day return policy and HOPEFULLY can see what AMD offers.
If AMD is good, return the 3080. If it's bad, open the 3080.

 

[DrHudacris]

I'm going to attempt to buy a card tomorrow BUT I won't be opening it. I'll attempt to buy a 2nd card in mid-October (return the launch day card) to take advantage of the 30-day return policy and HOPEFULLY can see what AMD offers.
If AMD is good, return the 3080. If it's bad, open the 3080.

My plan too!

Edit: but I might give in to my weakness and open it...I'm sure with the scarcity in supply I can resell without much loss

 

[DmanX]

FLIR imaging of RTX 3080 (from Guru3D): Says to me that this will work in the T1 with PSU offset and (2)120mm x 15/25mm Exhaust fans on top. It's hot on the pass-through side, relatively cool on the rear exhaust side.

Picture orientation reference:

URL: https://www.guru3d.com/articles_pages/geforce_rtx_3080_founder_review,9.html

 

[ghostfish]

I'm looking at the teardown of the 3080 and it looks like the backplate will be required since VRMs are found on the back of the PCB.

Those are just capacitors on either side of the chip, they don't produce much heat and there are thermal vias all around them on the PCB. They don't need cooling. The only thing you may have to worry about is WPM controllers near the V notch that have thermal pads above them on the stock backplate. I don't think they're going to produce much heat though. It's just a controller/gate drive, the FETs are on the top side. Most likely back side thermals aren't going to be a problem.

FLIR imaging of RTX 3080 (from Guru3D): Says to me that this will work in the T1 with PSU offset and (2)120mm x 15/25mm Exhaust fans on top. It's hot on the pass-through side, relatively cool on the rear exhaust side.

In these photos on the edges you're seeing housing temperatures, not fin temperatures. Similarly, on the bottom of the GPU on the flow through side you're seeing the fin temperatures directly, but on the back exhaust side you're seeing the temperature of the intake fan and the shroud around it, not the fins. What we really want to know is what the air temperature is coming out of each side under load.

 

[Luke NK]

The 3090 should fit right? Sadly if it does there won't be space to offset the PSU and leave room for the flow through fan to exhaust hot air

 

[DmanX]

In these photos on the edges you're seeing housing temperatures, not fin temperatures. Similarly, on the bottom of the GPU on the flow through side you're seeing the fin temperatures directly, but on the back exhaust side you're seeing the temperature of the intake fan and the shroud around it, not the fins. What we really want to know is what the air temperature is coming out of each side under load.

For my best guess, I would say the air temps are going to be much lower than the fin temps. I'd imagine there would be a halo protruding from the fans representing the movement of hot air under a FLIR image if the air temp was hot enough to register or it is registering in the top shot at a temp of <37 degC. I'm referring to the very light blue halo in the area in the first photo coming from the pass-through end of the GPU. That should represent ambient air flow (bowing gradient of light blue).

 

[ghostfish]

For my best guess, I would say the air temps are going to be much lower than the fin temps. I'd imagine there would be a halo protruding from the fans representing the movement of hot air under a FLIR image if the air temp was hot enough to register or it is registering in the top shot at a temp of <37 degC. I'm referring to the very light blue halo in the area in the first photo coming from the pass-through end of the GPU. That should represent ambient air flow (bowing gradient of light blue).

Thermal cameras measure the IR radiation coming off a surface (+ reflected + transmitted, though these are usually small). They're more or less unaffected by air conditions. You're seeing the temperature of the surface in that image, not the air. They are impacted by the emissivity of the material, although for most painted non-reflective metals and plastics it's pretty consistent around 0.95.

The most likely reason for "smearing" in that image is they're using a relatively cheap thermal camera. High resolution thermal cameras are very expensive, so what most of the models <$1,000 do is have 2 cameras. 1 high resolution visible light camera, and 1 low resolution thermal camera. They overlay the high res visible light image over the coarse thermal data to make the thermal image look higher resolution than it is.