Resolved Sapphire PULSE RX580 8GB & Dual Mini 6 Pin to 8 Pin PCI-E Power Adapter: 16AWG or 18AWG?

[koreda]

When I purchased my new Sapphire 11265-05-20G Radeon Pulse RX 580 8GB GDDR5 Dual HDMI/ DVI-D/ Dual DP OC W/BP (UEFI) LITE (from Newegg) I only made a cursory investigation into the necessary Dual Mini 6 Pin to 8 Pin PCI-E Power Adapter required to power the GPU.

Now that the card is kaput after only 6 weeks of use (w no visible errors), I'm wondering if I erred when I chose the 18AWG over the (more popular? common?) 16AWG? Looking back I originally went with 18AWG bc of it's Rated Power: 300W [6.5A], which is also the macpro5,1 'combined max output (75X x4) for all PCI Express slots.' Only after looking for lower prices did I discover the 16AWG adapter, but ultimately went along with "Amazon Choice for sapphire radeon pulse rx 580 8gb gddr5..." based on price, delivery and user comments.

Is this even accurate? Searching through comments here, I now see the 16AWG is mentioned more frequently, especially with this mac pro + GPU.

Could anyone offer any insight here; Could the wrong wire gauge could in fact lead to GPU failure? Should I pitch the 18AWG adapter and only use 16AWG? I only assumed it was just a matter of insulation but know I see:​

• Gauge: 18AWG, Rated power: 300W; [6.5A]
• Gauge: 16AWG, Rated power: 450W; [10A]

Since I'm utilizing the combined 300W from each mini 6-pin + 75W yielded from PCI-E 16x graphics Slot #1, should I have in fact gone with the 16AWG (w it's 450W rated power) pie power adapter? If I need more power, i.e. on-board SATA, is there another power adapter I should be looking at?

=============================================================
EDIT:
Radeon RX 580 Requirements:
Typical Board Power (Desktop) PSU Recommendation
185 W 500 W
The supporting documents and media provided with all new Sapphire Pulse GPUs is staggeringly sparse. I was able to cull the initial 500W PSU Recommendation from Sapphire's website, but only AMD (website) would yield the distinction, specifically addressing the problem included in this post; Albeit; Sapphire does list: 'Power Consumption < 225W'

EDIT II:
The gauge of the power adapter had little (to none) effect with the failed GPU [confirmed, resolved - thanks for that!]
Nevertheless, considering the approximate maximum current draw of the mobo PCI-E Slot(s) and Aux (375W), and the four (4) cards currently installed in my cmp5,1 configuration, it only follows to utilize the corresponding PCI Express power cable, even though the noticeable difference will be negligible.
When I asked COMeap about possibly exchanging the 18AWG for the 16AWG, they immediately replied with an offer of a new 16AWG adapter, free of charge! Considering the stalled-mire of Luddite customer service I'm currently experiencing with (processing RMA) Sapphire and their third-party "warranty-replacement" outfit Althon Micro Inc, replacing my new Sapphire Pulse GPU, it's good to know caliber merchandise and genuine customer care still exists beyond Apple.
=============================================================

These were the power adapters I originally considered:
COMeap Dual Mini 6 Pin to 8 Pin PCI Express Video Card Power Adapter Cable for Mac Pro Tower 15-inch [Amazon, $13] (currently installed)

COMeap 16AWG Dual Mini 6 Pin to 8 Pin PCI Express Video Card Power Adapter Cable for Mac Pro Tower 15-inch [Amazon, $16]

OEM Apple Dual Mini PCI-E 6-Pin to Standard PCI-E 8-Pin Video Card Cable [$15]

Additional info:
Part Number [pn]: 299-4E353-830FA. Unflashed (PC version) worked out of the box. [for six weeks].

FileVault inactive, Errday. macOS High Sierra 10.13.6 | 2 x 3.46 GHz 12-Core Intel Xeon | 128 GB 1333 MHz DDR3 |
PCI-E 16x Lane #1: ATI Radeon HD 5770 (archive GPU) |
PCI-E 16x Lane #2: Apricorn Velocity Solo x2 Extreme w/ Crucial MX300 3D TLC SSD |
PCI-E 8x Lane Slot #3: MAXPower RAID Mini-SAS 6G-1e1i Controller w/ 4 x WD Black 2TB HDD |
PCI-E 8x Lane Slot #4: Sonnet Allegro USB 3.0 4-Port PCI Express

 

[iluvmacs99]

I don't think the cables caused your Radeon to die. Different gauges meant that it could take the load amperage before it melt the surrounding casing which can cause a fire hazard. Preferably, the 16AWG is better, but 18AWG should work fine as the Radeon Pulse version draws no more than 225W (75Wx3) during operation. Mine is the 18AWG version. If it draws more, then your Mac Pro would have shut down anyhow, which is why some people need the Pixlas mod to provide more power to power hungry cards. GPU cards do fail and with AMD, they offer a 30 days no question asked full warranty replacement on dead cards and the remainder of the 3 year warranty mailed in. So if you would had bought the card from an authorized store, which I did, you would have the card replaced. What is more important is "HEAT" and airflow inside the machine. I placed my Radeon on slot 2 which gives more airflow space for the fans to dissipate heat and for the Mac Pro fans to blow cold air directly to the card. I place my GT 120 card on slot 1 as it is a smaller profile.

 

[h9826790]

Most of your numbers or assumptions are not correct.

1) the GPU won't fail because you used a "wrong" Gauge cable. The GPU won't care what cable you are using. It will just pull whatever it need. In worst case, the cable isn't strong enough, and catch fire. The cable may fail, but the GPU will not (because of the Gauge).

2) the CMP has 980W PSU, not "can only put out < 400W".

3) that 500W PSU requirement is really PSU rating, not how much it can output to the GPU (no matter via slots of aux power cable)

4) the PULSE RX580 8GB only draw ~225W max. Both the cMP and your cables can easily handle that.

If the graphic card failed in this short period of time, you better use the warranty.

 

[koreda]

Most of your numbers or assumptions are not correct.

1) the GPU won't fail because you used a "wrong" Gauge cable. The GPU won't care what cable you are using. It will just pull whatever it need. In worst case, the cable isn't strong enough, and catch fire. The cable may fail, but the GPU will not (because of the Gauge).

2) the CMP has 980W PSU, not "can only put out < 400W".

3) that 500W PSU requirement is really PSU rating, not how much it can output to the GPU (no matter via slots of aux power cable)

4) the PULSE RX580 8GB only draw ~225W max. Both the cMP and your cables can easily handle that.

If the graphic card failed in this short period of time, you better use the warranty.

everyone needs an editor

 

[startergo]

Should I pitch the 18AWG adapter and only use 16AWG?

The wire thickness is only gonna make difference to the "eventual fire"
i.e. it's like the water hose thickness:
How much current/water can you pass through that diameter? The thicker the wire/hose the bigger the current/flow.

 

[crjackson2134]

OP, my brand new RX580 failed after a few weeks too. I did an RMA and was sent a refurbished unit. It’s been performing flawlessly ever since. Your issue has nothing to do with the gauge of your power cables, and probably everything to do with quality control. Just make them replace it, and move forward.

 

[koreda]

It seems like the rate of (early) failure and RMA process with this 500 series isn't discussed enough.

I already initiated RMA almost two weeks ago. It's a stop and go process, they initially didn't reply for a week, until I submitted another ticket.
Now I'm at the mercy of Althon Micro and am well aware of their bottomless well of refurbished GPUs, e.g. my eventually, new'ish RX 580 card, once they resume communications again (after 64-hrs radio silence again)

Rather than go through this entire time-wasted process again I'd hope to re-trace my install steps, before committing to the exact same installation system and GPU operation that may have led me here in the first place. I'd hoped to gain some insight, but I obviously committed to the wrong question (albeit, I do appreciate the confirmation).
Unfortunately, from what I've read at AMD forum and reddit, it's highly probably this will happen again (sometime before Sapphire's parts warranty lapses).

I can't afford not to have a high(er) end GPU installed on my remote workstation (home office), as I absolutely need it for finishing strategies.
The no nonsense solution will likely be a backup (XFX Radeon RX 580) GPU for when this Sapphire Pulse fails again and run the GPU off a separate PSU outside of the tower, a more economical solution to a Pixlas mod. (I leave the side panel off for ventilation, so cabling want be an issue).
Of course 'just make them replace it' will happen first, hopefully[?]

 

[h9826790]

I leave the side panel off for ventilation, so cabling want be an issue

For info, this will destroy the designed airflow. The open fan type graphic card may have better cooling. But everything else will have worse cooling.

Since the graphic card has its own fan and sensors. It's better to keep the side panel close for the cooling of logicboard / RAM / etc.

https://forums.macrumors.com/thread...-dissipate-better-heat.2130514/#post-26315403

 

[crjackson2134]

It seems like the rate of (early) failure and RMA process with this 500 series isn't discussed enough.

I already initiated RMA almost two weeks ago. It's a stop and go process, they initially didn't reply for a week, until I submitted another ticket.
Now I'm at the mercy of Althon Micro and am well aware of their bottomless well of refurbished GPUs, e.g. my eventually, new'ish RX 580 card, once they resume communications again (after 64-hrs radio silence again)

Rather than go through this entire time-wasted process again I'd hope to re-trace my install steps, before committing to the exact same installation system and GPU operation that may have led me here in the first place. I'd hoped to gain some insight, but I obviously committed to the wrong question (albeit, I do appreciate the confirmation).
Unfortunately, from what I've read at AMD forum and reddit, it's highly probably this will happen again (sometime before Sapphire's parts warranty lapses).

I can't afford not to have a high(er) end GPU installed on my remote workstation (home office), as I absolutely need it for finishing strategies.
The no nonsense solution will likely be a backup (XFX Radeon RX 580) GPU for when this Sapphire Pulse fails again and run the GPU off a separate PSU outside of the tower, a more economical solution to a Pixlas mod. (I leave the side panel off for ventilation, so cabling want be an issue).
Of course 'just make them replace it' will happen first, hopefully[?]

I agree that the RMA process could be a little better. I was a little put off by the cosmetic condition of my replacement GPU at first. That said, I quickly got over that since the replacement worked so much better. I too had to make repeated requests to Alton Micro to kick the ball into play. Initial communication was with their primary intake tech, and he was very responsive. Then I was directed to a second person who actually processed the RMA. The second contact was slow and unreliable. I had to keep prodding him to get this done. It wasn’t a horrifically long process, but it was by far the slowest RMA I’ve ever had. I give them 3 out of 5 stars. I believe it took about 3 weeks IIRC.

If your use of this GPU is critical, I recommend installing it in PCIe 2 to give best possible airflow space around the card, keep the side panel ON and having a tested/proven backup GPU on hand just in case.

 

[1madman1]

• Gauge: 18AWG, Rated power: 300W; [6.5A]
• Gauge: 16AWG, Rated power: 450W; [10A]

Those ratings are PER CONDUCTOR, 18AWG is plenty. Remember, the Apple supplied 2 conductor power cord leading from the mains to your computer is probably 16AWG and it has to power everything in your computer combined. It's very common for PC desktops to only use 18AWG mains power cables, including the HP Z420 I have sitting next to me.

Even if the thickness of the wiring was too small, it still wouldn't kill the card. It just means the wiring would run hotter than normal under load.

 

[orph]

some times part's fail, from what i know most parts die in the first 8 weeks then around after 2-3 years (depends on what it is)

so i think you just got bad luck

thats what RMA is for, if you have the option (in the UK you can) return it to the shop for a refund and buy a new one or swap in the shop.
amazon tends to be vary good (for the first year amazon has to swap it in the UK i think?) they will just send out a second unit once they get yours.
or some shops in the uk are well know for looking after customers like https://www.overclockers.co.uk/