Custom R9 290/290X Reviews/Availability/Listings thread

[Fastx]

Re: Sapphire 290 Tri-X
I get 77C/86C/57C Core/VRM1/VRM2 under maximum load scrypt mining, OC'd with an ambient of 27C/~81F. That's far above a normal gaming load. You would certainly need more exotic cooling for a 24/7 overvolt + overclock, but it's limited by its reference PCB anyways. If your application is gaming the Tri-X is one of the better stock coolers I've tested. A Sapphire @ 2000 RPM is comparable to a Gelid Icy Vision @ 4200 RPM in cooling performance while being a quiet whoosh compared to the Gelid's roar.

Re: MSI 290 Gamer
We'll see when I get my MSIs. Supposedly also based off a reference PCB with a few component swap outs (coils + RAM - supposedly Hynix instead of Elpida).

I'll take a closer look at the components and will be sure to post on cooling/component choices/OC-ability.

IEC what is the overclock core clock, memory overclock clock and voltage on the above in bold?
(TIA)

Here is a link below for that user/review on the MSI 290 Gaming card fwiw. I had a hard time going back and finding it again, but its only one user/review and one card but fwiw you can check it out per link below. The members name is steady2004, this could just be a bad card and it is only one review so we will see how the card does for you and hopefully better. If I were to get a AIB 290 there is no way I would ever buy one of these cards without VRM temp sensors for overclocking/ gaming imo.

http://www.overclock.net/t/1453703/vr-zone-msi-launches-r9-290-and-r9-290x-gaming-4g-video-cards/30

 

[24601]

IEC what is the overclock core clock, memory overclock clock and voltage on the above in bold?
(TIA)

Here is a link below for that user/review on the MSI 290 Gaming card fwiw. I had a hard time going back and finding it again, but its only one user/review and one card but fwiw you can check it out per link below. The members name is steady2004, this could just be a bad card and it is only one review so we will see how the card does for you and hopefully better. If I were to get a AIB 290 there is no way I would ever buy one of these cards without VRM temp sensors for overclocking/ gaming imo.

http://www.overclock.net/t/1453703/vr-zone-msi-launches-r9-290-and-r9-290x-gaming-4g-video-cards/30

The Chip gimping has begun <.<

Any AIB dumb enough to not have VRM sensors on this card should not get money.

This ain't a GK110 card with little VRM cooling problems, this is a Hawaii card with massive VRM cooling problems.......

@IEC, can you give me your clocks/voltages and hash rate for ~3 hours of scrypt mining @100% fan with Core/VRM1/VRM2 temps so I can compare our numbers? Thanks.

PCIe Slot 0-1:XFX Reference 290 72.7% ASIC Quality 1000 core@-0.00625v core 1500 ram @ 877.8 kh/s
PCIe Slot 3-4:XFX Reference 290 68.1 ASIC Quality 1000 core@+0.00625v core +0.00625v aux 1500 ram @ 877.8 kh/s
PCIe Slot 6-7:XFX Reference 290x 71.9% ASIC Quality 1025 core@+0.0125v core +0.0125v aux 1500 ram @ 983.4 kh/s

ASUS PT1 BIOS
Ambient Temp: 83F
PCIe Slot 0-1:290(1): VRM1/VRM2: 64C/71C GPU-z Wattage: ~223.3w
PCIe Slot 3-4:290(2): VRM1/VRM2: 61C/67C GPU-z Wattage: ~217.5w
PCIe Slot 6-7:290x: VRM1/VRM2: 64C/68C GPU-z Wattage: ~250.3w

Can't get the GPU Core temps without crashing the cards. Managed to stop all the cards from throttling now though.

 

[Silverforce11]

The Chip gimping has begun <.<

Any AIB dumb enough to not have VRM sensors on this card should not get money.

This ain't a GK110 card with little VRM cooling problems, this is a Hawaii card with massive VRM cooling problems.......

@IEC, can you give me your clocks and hash rate for ~3 hours of scrypt mining @100% fan with Core/VRM1/VRM2 temps so I can compare out numbers? Thanks.

I thought a lot of the 780 ti cards run their VRMs at 85-90C in gaming loads .. as per thermal imaging from Guru and hardware.fr, good on NV for not including any VRM sensors, nobody needs to know!

http://forums.anandtech.com/showpost.php?p=35877711&postcount=101

 

[24601]

I thought a lot of the 780 ti cards run their VRMs at 85-90C in gaming loads .. as per thermal imaging from Guru and hardware.fr, good on NV for not including any VRM sensors, nobody needs to know!

http://forums.anandtech.com/showpost.php?p=35877711&postcount=101

Nvidia sets conservatively high stock voltages for their chips. AMDon't. AMD sets their voltages at the very edge of stability and therefore is heavily effected by VRM temperatures as you can only really go up to 120C on VRMs and many AMD AIBs like to push that up to 105c @ stock volts.

Higher VRM temps = higher ripple, making that which was stable before not stable now.

Needing to increase voltage to make up for that deficiency on AMD AIB cards quickly leads to thermal runaway in the VRM area.

TL: DR, NVIDIA sets high stock voltages compared to what their GPUs need. AMDon't.

 

[blackened23]

Faulty VRM cooling would indicate higher RMAs associated with a particular brand. I wonder, does anyone have concrete historical data points, such as, RMA rates for AMD cards versus RMA rates for nvidia cards in 2013? Has anyone found data for RMA/failure rates of AMD AIBs vs nvidia AIBs for the year 2012 or the year 2013? Just curious. I'm sure nvidia's VRM cooling would have a role to play with that data. Clearly some indicate in this thread that nvidia has VRM problems. Or certain 290X aftermarket models have VRM issues. I'm sure that could play a role with concrete data points.

You can't argue with concrete data points. Or can you? Welp i'm off to searching for concrete data. I'll be back in a bit.

 

[wand3r3r]

^ good luck with that, I'm sure they publish them.

My 290x stock VRMs were very low, 70's iirc mining. (stock core, 55 or was it 65% fan when I checked I have to double check, 1500 memory)

 

[blackened23]

^ good luck with that, I'm sure they publish them.

My 290x stock VRMs were very low, 70's iirc mining.

AIB failure rates aren't published? Challenge accepted. I'll be back in a bit.

 

[24601]

It doesn't change RMA rates one bit, as you can RMA infinite times and get the exact same design.

What it changes is Return rates.

 

[blackened23]

So what's the most concrete data point to look for? I'm assuming that VRM cooling failure would lead to one of the following: Failure, RMA or return. Or all three.

Failure rate? RMA rate? Or return rate? And we'll see how this ties in with AMD AIBs vs nvidia AIBs, and maybe that can correlate to VRM cooling.

 

[24601]

I see you didn't understand my post about VRM Temps vs Voltages vs Needed Voltages vs Voltage buffer in relation to needed voltage.

If your stock voltage has a high amount of buffer between needed voltage and stock voltage, your VRMs are fine until VRM failure point, which is about 120+C.

If your stock voltage has a very little amount of buffer between needed voltage and stock voltage, your VRM's nominal temperature determines Input voltage ripple, which determines actual minimum input voltage due to that ripple. If this ripple reaches a large enough amount, the stock voltage becomes lower than needed voltage and you have crashing going on.

From everything I've seen, AIB cooling is usually designed only around core temperature (if that) and cost. VRM temperatures are only tested with low intensity workloads if at all, and even then, the AIBs seem content with 110C-120C as their target point for VRM cooling.

For Nvidia cards they can get away with this as a result of what I discussed earlier. For AMD cards, there's no way in hell they can get away with this, as seen by my previous discussion of AIB ad infinitum.

Both AMD and Nvidia have used "Powertune" and the like to try to limit their required TDP dissipation budget as well as component ratings as well as the all-important PCIe standards for their stock cards.

The AIBs take advantage of that by just making use of the throttling to have to do less for their cooling paradigm, core, VRM, and memory all accounted for.

This lie about "power viruses" is the rabbit hole we went down and now we cannot dig our way out of it, or even acknowledge that we are in the hole.

 

[IEC]

IEC what is the overclock core clock, memory overclock clock and voltage on the above in bold?
(TIA)

Here is a link below for that user/review on the MSI 290 Gaming card fwiw. I had a hard time going back and finding it again, but its only one user/review and one card but fwiw you can check it out per link below. The members name is steady2004, this could just be a bad card and it is only one review so we will see how the card does for you and hopefully better. If I were to get a AIB 290 there is no way I would ever buy one of these cards without VRM temp sensors for overclocking/ gaming imo.

http://www.overclock.net/t/1453703/vr-zone-msi-launches-r9-290-and-r9-290x-gaming-4g-video-cards/30

1050/1475 stock voltage, powertune + 20%, gets over 910 kH/s.

Edit: Fan 44% ~2200RPM

I could run the fan higher and get nicer temps but it's nice and quiet.

 

[24601]

1050/1475 stock voltage, powertune + 20%, gets over 910 kH/s.

Thanks; what fan speed in % was that from? 100%?

If that was @ 100% fan, at least that means I can finally write off all the AIB coolers for scrypt mining.

86C vrm1 @ 1050 @ stock voltage does not inspire confidence.

 

[Fastx]

Right now the MSI 290 Gaming TF IV is $469.99 shipped at Amazon if anyone is interested but make sure you read the user review in post 301 first fwiw.

http://www.amazon.com/gp/product/B00...?tag=anan06-20

 

[blackened23]

Out of stock and usually ships in 1-2 months? And that's 60$ over MSRP isn't it?

What if someone wants to mine now , not two months from now.

 

[Fastx]

Out of stock and usually ships in 1-2 months? And that's 60$ over MSRP isn't it?

What if someone wants to mine now , not two months from now.

The first time I seen it I didn't see that ship in 1-2 months, but it is $110.00 cheaper than any other AIB 290 and yes the reference was I believe was around $399.00 when released.

 

[24601]

Right now the MSI 290 Gaming TF IV is $469.99 shipped at Amazon if anyone is interested but make sure you read the user review in post 301 first fwiw.

http://www.amazon.com/gp/product/B00...?tag=anan06-20

There's no way prices will stay up for 2 months.

Newegg already had to cut the reference back down to 500 USD for 290 and 600 USD for 290x.

For a card like 290/x I would definitely wait for the lightning if you are an extreme overclocker anyways, which will likely be out by those 2 months.

 

[Fastx]

There's no way prices will stay up for 2 months.

Newegg already had to cut the reference back down to 500 USD for 290 and 600 USD for 290x.

For a card like 290/x I would definitely wait for the lightning if you are an extreme overclocker anyways, which will likely be out by those 2 months.

That is my plan to wait a couple months and see where prices go and Mantle performance. I wondering if Mantle lives up to the performance if this will keep a premium price on the 290's. I am not saying it will but this has crossed my mind.

 

[blackened23]

I wonder if BF4 was the best showcase for Mantle? The sales figures have been rather disappointing and the player base for BF4 has been disgusted by the buggy nature of the game. Not just on PC either - it's buggy on the XB1 and PS4 as well for some reason.

Just thinking out loud here. Maybe COD Ghosts would have been the best Mantle show case game? COD Ghosts is outselling BF4 6 to 1 at last check and AMD probably could have thrown in some magic to make the graphics better with Mantle. And not deal with DICE's crap in the process.

But yeah. I could see how a show case title would help AMD sell their custom 290/X cards. Definitely agree with you there Fastx. I don't think BF4 would have been the best title in retrospect, though. Clearly the playerbase is pretty disgusted by DICE right now, and the sales show it.

 

[Skurge]

I wonder if BF4 was the best showcase for Mantle? The sales figures have been rather disappointing and the player base for BF4 has been disgusted by the buggy nature of the game. Not just on PC either - it's buggy on the XB1 and PS4 as well for some reason.

Just thinking out loud here. Maybe COD Ghosts would have been the best Mantle show case game? COD Ghosts is outselling BF4 6 to 1 at last check and AMD probably could have thrown in some magic to make the graphics better with Mantle. And not deal with DICE's crap in the process.

But yeah. I could see how a show case title would help AMD sell their custom 290/X cards. Definitely agree with you there Fastx. I don't think BF4 would have been the best title in retrospect, though. Clearly the playerbase is pretty disgusted by DICE right now, and the sales show it.

Are you suggesting COD would be a better showcase for Mantle than BF4? As buggy as BF4 is, I can't imagine the is a better game to make good use of Mantle. Large maps, huge numbers of players, heavy on the physics.

 

[Rvenger]

Right now the MSI 290 Gaming TF IV is $469.99 shipped at Amazon if anyone is interested but make sure you read the user review in post 301 first fwiw.

http://www.amazon.com/gp/product/B00...?tag=anan06-20

According to OCN this card has worse cooling than the reference cooler. No VRM temp readings (If the GTX 780 is high, these have to be higher) It also uses chokes inferior to the reference design. Magictech vs bussman. My advice would be to stay clear of this card.

 

[24601]

According to OCN this card has worse cooling than the reference cooler. No VRM temp readings (If the GTX 780 is high, these have to be higher) It also uses chokes inferior to the reference design. Magictech vs bussman. My advice would be to stay clear of this card.

I'm just confused how they failed so badly considering Gigabyte used the exact same concept for their VRM cooling (Direct contact heatsink fins connecting the VRMs directly to the main heatsink) and did very well (in AIB cooling terms) with it.

http://www.guru3d.com/articles_pages/gigabyte_radeon_r9_290x_windforce_3x_oc_review,12.html

87.3C @40% fan
75.6C @55% fan

Was it because of the tantalum capacitors on the Gigabyte cards along with what looks like the same chokes as on the Nvidia GK110 cards?

 

[thilanliyan]

TL: DR, NVIDIA sets high stock voltages compared to what their GPUs need. AMDon't.

Not true in all cases I don't think. All 3 of my 7950s for example run at 0.975v at stock clocks. Stock voltage is 1.25v which is way higher than what's needed. And I'm pretty sure I've read of others with similar thoughts in the crypto thread that say that 1.25v for a 7950 is way higher than what's required.

 

[24601]

Not true in all cases I don't think. All 3 of my 7950s for example run at 0.975v at stock clocks. Stock voltage is 1.25v which is way higher than what's needed. And I'm pretty sure I've read of others with similar thoughts in the crypto thread that say that 1.25v for a 7950 is way higher than what's required.

The shoehorned in boost for the 7950s was very badly done.

I'm thinking most of the reason for the crazy headroom given in that case was to make room for many AIBs to cut crazy corners in component quality (which they did)

It also allowed some very wide binning for the chips that went into 7950s, as I know that many of them weren't able to achieve the feat you describe.

Only 70%+ ASIC Quality chips were able to mine at 0.975v at stock clocks to my knowledge.

 

[Fastx]

According to OCN this card has worse cooling than the reference cooler. No VRM temp readings (If the GTX 780 is high, these have to be higher) It also uses chokes inferior to the reference design. Magictech vs bussman. My advice would be to stay clear of this card.

Yeah I was hoping when the 290 MSI Gamer was released it would perform good. I have this cooler on my 770 and it is a very nice/quiet cooler at all loads which is what I like imo. It's too bad so far, but look to see what IEC thinks when he gets his as it is only one review, but no VRM's temp sensors is a deal breaker regardless for me on any 290 imo.

 

[blackened23]

According to OCN this card has worse cooling than the reference cooler. No VRM temp readings (If the GTX 780 is high, these have to be higher) It also uses chokes inferior to the reference design. Magictech vs bussman. My advice would be to stay clear of this card.

I don't really get it. These coolers perform superbly on the nvidia side (MSI gamer and the Asus DC II) yet when they slap them on AMD cards, MSI and asus seem to treat the AMD side as a red-headed step child.

Not sure what's up with that. Anyone know? There seems to be a rather big gulf with these coolers being excellent on the NV side, not so great on the AMD side. Although i'm not sure if that is solely due to VRM cooling or what. I suspect that the Nvidia green light program plays a role in ensuring things are up to snuff, but I'd think that AMD has something similar in terms of ensuring AIB quality. Especially after the debacle with the XFX DD 79xx cards.

Right now it seems the Tri-X is the best one on the AMD side, someone correct me if i'm wrong. From what i've seen the Tri-X 290X performs hands down excellent even with VRM cooling.

In the end, we should demand better of these hardware companies. I don't think it's acceptable that both the Asus and MSI aftermarket boards have sub-par VRM cooling. It's pretty anti-consumer, IMO, and i'm not sure what the best method of sending a message to asus and MSI is. I guess vote with your wallet and get the Sapphire Tri-X over those cards.