All you need to know about quality of MSI hardware

This company is so cheap that they use undersized thermal pads on their video card.

 

Look at this GTX 1660 Ti from Asus.
https://bit-tech.net/reviews/tech/graphics/asus-geforce-gtx-1660-ti-phoenix-oc-review/1/
No thermal pads on the VRM's !!

 

Higher wattage cards in higher-end SKUs than the GTX 1660 Ti Phoenix have also been shipped with bare VRMs. The Super Alloy Power II design is more efficient than reference design, thus producing less waste heat. Official numbers are up to 35°C lower temperature and 2.5X extended lifespan. Redundancy aside, putting a thermal pad over the VRMs of the Phoenix would, from the looks of it, impede exhaust airflow. What I'm saying is that I doubt Asus was cheap or careless in their omission.

 

https://bit-tech.net/reviews/tech/graphics/asus-geforce-gtx-1660-ti-phoenix-oc-review/12/

Unfortunately, the cooler really is rather poor. A delta T of 60°C is not a good result for a GTX 1660 Ti, with Palit and Zotac managing 46°C and 50°C respectively with similarly sized cards. The card also reports that its boost speed is being thermally limited (and power-limited), which is not a good sign. Asus’ card is also on the loud side compared to the others.

In OC Mode, the card boosts to 1,830MHz (still lower than the Palit stock card) and is still reported as thermally limited. This’ll get you 1-2 percent more performance going by 3DMark Time Spy, but the fan gets noticeably more noisy in its attempts to cope, and the temperature goes up by 3°C too

Seems the VRM's are over-heating and thermally limiting the overclock.

For comparison, look at the Palit GTX 1660 Ti
https://www.guru3d.com/articles_pages/palit_geforce_gtx_1660_ti_stormx_review,5.html

 

I'm a bit confused by one of bit-tech's statements:

The memory dies appear to be making contact with the heatsink to me.


Anyhow, I don't wish to be quarrelsome, and I realise that the Phoenix is a card which re-uses a cooler from Asus' parts bin, adding an SKU to their range at a minimal development cost for a diversified portfolio. As reviews show, a GTX 1660 Ti is powerful enough to warrant a larger heatsink than what is provided by the Phoenix, and there are better alternatives out there. My conjecture was directed at the VRM cooling only. I remain doubtful the card would benefit from a thermal pad running across the VRMs, if the rest of the cooler remained unchanged.

Because the Phoenix isn't set up to be an OC champ, having only limited potential to handle additional heat coming from the GPU die, I think there are limiting factors being reached too soon for the VRMs to reach an alarming temperature and impact performance. The boost speed being power-limited doesn't necessarily stem from overheating VRMs, and since there's a thermal limit as well, adding a slight restriction to the exhaust airflow in the form of a thermal pad seems detrimental rather than beneficiary. I could of course be wrong and it would be interesting to see a thermal image of the PCB under load.

 

Did you watch the MSI video? It seems that cooler is also re-used.

youtu.be/morJq0HJoCc?t=739

Does this look like an overheating processor to you?
https://bit-tech.net/reviews/tech/graphics/asus-geforce-gtx-1660-ti-phoenix-oc-review/10/
If the graphics processor is not causing thermal throttling, then what is? The vram is fully covered by thermal pads and makes contact with the heatsink.

I used to own an Asus GTX 750 Ti and it had a crappy cooler too. No thermal pads on both the vram and vrm's.
It upsets me that you are so eager to defend Asus. Be honest. What is your motive here?

 

Yes, I did. No excuse not to use full-size thermal pads, and the cooler design does indeed look familiar.

Those two slides show a power draw of 265W and a temperature of 63°C while playing BF4. Unless I'm missing something, those values are healthy for a 1660 Ti. I find them indicative of neither thermal nor power throttling.

If I'm not mistaken, the throttling you referred to earlier was to the boost clock. Power limit, voltage, vBIOS, binning, and temperatures still below junction are alternative culprits. Can't rule out the VRMs overheating either, but it's not the only possible explanation and, in my opinion, it's not the most likely one.

I still have an old Asus 290X DCII OC around somewhere, and I can't say I regret that purschase as it was notably quieter than reference, had a nice backplate, and barely commanded a premium over reference. Just as you discovered on your 750 Ti, however, there were no thermal pads on the VRMs or memory modules. I'm not sure if memory overclocking was bad because of low quality VRAM or lack of cooling, but it would have been nice if they hadn't skimped on pads. I find it less acceptable to not properly cool all hotspots of a card if it has a beefy cooler and is advertised as overclocking-friendly.

I'm not eager to defend Asus, and I don't have any other motive than a desire to voice my opinion. I'm simply not convinced a VRM thermal pad would have been an improvement to the Phoenix, given how it would block off a small amount airflow and we don't know what temperature the VRMs are reaching without it. I'm perfectly happy to tear that card apart in other respects, such as the small heatsink, inconvenient 8-pin power connector location, and its less than generous 110% power limit. I currently use a card from EVGA, and have had cards from XFX, MSI, and Sapphire as well. I'm fairly brand agnostic.

 

The Asus Phoenix GTX 1660 Ti has no heatpipes and no direct sinked cooling for the VRM's.
We should be condemning Asus for inadequate cooling, not spending time arguing that thermal pads on the vrm's won't help,
and that there is no proof that the vrm's are thermally throttling.

youtu.be/WmRfZez2oRk?t=1131

 

Its only a few pennies her and there but over many many GPUs it really adds up to millions saved.

 

I agree. A redesigned cooler which also provides VRM-cooling would have been much better. My point was that the current design doesn't accommodate a simple thermal pad fix to add proper VRM-cooling, unlike the MSI 5700 XT situation, where an appropriately sized heat plate is already in place. The Asus cooler is no doubt worse than the MSI, and even though it at least sticks to MSRP pricing, they lowered the bar too much. The EVGA 1660 Ti XC Black sets a different standard for MSRP. Way better cooler, as long as you're not case restricted to dual-slot.

 

Your initial reply strongly suggests that Asus VRM's do not overheat and thermals pads are generally unnecessary on Asus cards.
You are repeating the advertisement claims of Asus without any data to back it up.
Show me the data which proves that Asus VRM's do not overheat and are superior to VRM's that other manufacturers use.

 

My initial reply implies that Super Alloy Power II power components are more efficient than bargain bin components, so when used on a weak SKU GTX 1660 Ti card, they may not get as hot as you think. In no way did I suggest that Super Alloy Power II trumps top offerings from other manufactures, or that Asus' power design is generally better than others. The simple fact is that Super Alloy Power II is a nice upgrade to find on an MSRP card, and the Bit-Tech review you referred to states so as well:

In retrospect, I meant to bring up another possibility, not to dismiss the possibility that the card's power circuitry is indeed overheating, because we lack data. I haven't found a review yet which discusses VRM throttling or includes infrared images of this card. I haven't found anything for similar GTX 960 or 1060 SKUs either, but these budget-oriented, small cards aren't all that popular, I think. I know full and well cooling is usually needed, and that could be the case for this 120W TDP card too. It could be this card is showing minor throttling because it lacks cooling directed at the power circuitry specifically. I don't know and I was trying to head into this conversation open-minded.

I included that the numbers are "official", as in from Asus, as to indicate these are not independently measured numbers. I still included them to get the point across that more efficient VRMs can make a big difference to temperature. It's the same principle as more efficient PSUs needing less cooling. If I had found actual temperature measurements of the card's VRMs, I would have included those instead. I'm able to find as little data to back up my claim that the VRMs may not be overheating as you're able to find data backing up that they are overheating. The reason I replied to your original post in the first place was not to defend Asus, but to point out that even though the VRMs are not covered with a thermal pad, they may not run adversely hot. They could be, but it's not certain.

I also added how simply adding a thermal pad could slightly impair the cooler's airflow, to suggest that there could be a reason they didn't include one on this particular card. Yes, that means the cooler has a poor design, so no, that was not me praising Asus. If that actually is Asus' defense, it's still a cheap move since they didn't redesign their cooler to accommodate VRM cooling, but making a bespoke cooler for a low-margin, new card is a costlier endeavour than adding full-size thermal pads, I'd assume.

I have not suggested Asus is a superior manufacturer, nor have I suggested that no Asus card ever has overheating VRMs. The Strix Vega 64 is one example of an Asus card with uncomfortably hot VRMs, even though there actually is VRM cooling in place; it's just insufficient, with one VRM not properly covered. That's a 295W TDP card though, so not exactly in the same ballpark in terms of cooling needs. Neither of us have any data to once and for all settle whether the VRMs of the Phoenix do or don't overheat, but I sure wish we did, because I don't have an actual interest in the card itself or Asus' reputation. I found my alternative conjecture for the Phoenix not having a VRM thermal pad a valid addition to the conversation for the sake of discussing thermals only.

 

This reviewer says the Asus Phoenix starts throttling at 83C on the gpu.
youtu.be/mGTVhoOT8uI?t=118

How many graphics cards do you know that thermal throttle at around 85C on the gpu?
I never heard of a gpu which thermal throttles at 85C.

Later in that video, during the furmark test, the gpu is running at 89C.
How is it possible the gpu is running at 89C if the gpu thermally throttles at 83C?
Perhaps because the gpu is not thermally throttling? Something else is thermally throttling.

 

Pascal's BIOS hard limits GPU Boost 3.0 to 83°C. There's nothing in that video to suggest VRM-throttling.

Furmark is a stress test far more intensive than gaming. As you can see from the HWiNFO64 screenshot, the GPU clock has dipped to 1,489 MHz at 89°C, which means that not only has the boost clock been dialed all the way back, but because the GPU is still running hot, the card's been throttled slightly below stock core clock as well. Furmark is simply too much for the card to maintain an 83°C or lower core temp by using boost clock as a buffer.

On the subject of benchmarks, here's a video which shows only normal variation between the 1660 Ti XC Black and the 1660 Ti Phoenix: youtu.be/0bCt9592y-M?t=6m52s

The video also demonstrates how sample and benchmark variation is real, which I tried to highlight since the difference between Palit and Asus in Bit-Tech's review is too small to draw a conclusion. For the sake of defending VRM-overheating as a possible culprit, it's important to mention that the above benchmarks were run on an open-air test bench. It could be that the VRMs get too hot if the ambient temperature is high or the case the card is installed in has poor airflow. But again, it would be a case of which throttling point is reached first. In such a scenario, the GPU would run hotter too.

If it's okay with you, I'd prefer not keeping this particular conversation alive any longer since it all comes down to conjecture. We can revisit this topic if conclusive measurements appear.

 

When you can prove that bare VRM's on Asus cards do not over-heat, then I will stop arguing with you.

Why does the Asus ROG Strix have thermal pads on the mosfets, but the Asus Phoenix does not?
For cosmetic reasons?
youtu.be/RonKx7zr0bg?t=490

 

I've already clarified my stance and provided explanations, but you don't seem the least bit interested in the content of my replies. What you're doing now is rehashing your questions and accusations, and I will not engage in this conversation any further.

 

Thread is now closed. It's indeed been exhausted.

Charles