Yeah, looks too tight with that Fat^$$ GPU. I guess I’ll just have to forget about Slot 3.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
RX480 v. RX580 for Mojave
- Thread starter usna92
- Start date
- Sort by reaction score
Yeah, looks too tight with that Fat^$$ GPU. I guess I’ll just have to forget about Slot 3.
One of the things I really like about my Gigabyte GTX 1080 is that the top end of the card has a light labeled "Fan Stop" and the fans only work when needed. When the fans are working the "Fan Stop" Light (visible thru the grill) lets me know they are on. When they are not working the light is off. And, it's sort of cool, IMHO, that they change color.
In addition, there is another light that Says "Gigabyte" that is always on.
Lou
In addition, there is another light that Says "Gigabyte" that is always on.
Lou
I thought you might
The 7970 I had worked well for daily use and fit pretty well too, but there is a slight concern that lack of adequate air flow may have pushed it to early failure. It was an XFX card so it’s hard to say. They are known for early failure anyway.
I’d like to see the fit of a GTX 680 with the 7101A. The card is powerful enough for my daily needs in Mojave, but I’d need to see it. I won’t spend more money on an unknown fit. I can live w/o Slot3, but I don’t like it.
WX7100 won’t fit my personal wants. I’m hoping a slimmer alternative will come along. The HD5870 is a true double wide card, and it worked fine (fitement wise) just like the HD7970, but the cooling issue is a concern.
If someone would just buy my SSUBX so I can afford another 970 SSD for my raid card, it wouldn’t matter so much.
I’ll read up on it...
So after a bit of reading, I’m not sold on the card. It does offer single slot and lower tdp but for the price, I’ll just stick with the loss of PCIe Slot3 for now. Driver support does seem superior for pro creative apps, but I don’t need that. Everything else is par and sub-par to the 580 from what I’ve read. Even on the pro app support, most of what I read pertains to Windows and no mention of Metal 2 support (although I’m sure it’s there).
Last edited:
They are the same thing, just the chip / PCB production is a little bit optimised, and the BIOS give it a higher clock speed.
[doublepost=1536298917][/doublepost]
Enter this in terminal should able to check the GPU parameter
Code:
ioreg -l |grep \"PerformanceStatistics\" | cut -d '{' -f 2 | tr '|' ',' | tr -d '}' | tr ',' '\n'|grep 'Temp\|Fan\|%\|(W)\|Hz'Of course, it's your own decision. But in my experience, that "half open" fan for this kind of blower cooler is enough to let the whole cooling system to function properly. At least it's true on my 7950 Mac Edition card.
As long as the card in slot 2 won't touch the graphic card's fan, then it is OK.
Especially we can still lower the voltage / clock speed etc to make it run cooler. If we ever consider the WX7100 because of cooling. I will definitely pick this reference RX480 / 580 and run it with the WX7100 setting. From my RX580 VBIOS study thread, it shows a ~50W reduction by using the WX7100 figure to run the Sapphire PULSE RX580 8GB card. Even only half fan is opened, the double slot wide reference cooler will be way more than enough with this low power setting.
But of course, again, that's your decision, if you are not happy with that, no point to go for it. And not everyone happy to mod the VBIOS as well.
@h9826790 have you had the chance to perform the same VBIOS examination of the Radeon RX Vega 56? At a TDP of 210 watts, it theoretically fits within the available power to a cMP GPU: 75w from PCIe slot, 2x 75w from PCI power cables (total 225w), even if the reference version needs 6->8 pin PCI power conversion cables, however, it might be prudent to set its parameters more conservatively, and avoid the hopped-up (non-reference) gamer models that are ready for overclocking, overheating, and sucking down too much power for the cMP to deliver. It still boggles my mind that for less than $1,000 one can buy a > 10 TeraFLOP GPU - I worked for Apple during the 1990s in the group that maintained its series of Cray Supercomputers (X-MP/48 (800 MegaFLOPs!), then Y-MP/2E).
I ran one each of a Gigabyte and an XFX reference model Vega 56 for a bit under 10.13.6 but neither seemed to be doing good enough thermal management and I had system freezes (I have no idea what VBIOS parameters they have because I eBay'd them both (used)), so I've backed down to a nice, cool, power-efficient RX 480 for now (it still beats my old flashed HD 7950 Boost, and can apparently run 2x 4K monitors without too much fuss, though it's very annoying to lose EFI boot screens). I really wish I didn't have to install Windows to poke at them, but ...
I am very, very happy to see 5GT/s PCIe 2 bandwidth being correctly negotiated (and checked via CL!ng) now that I've installed the 138 EFI boot ROM update from Mojave Beta (still running 10.13.6 with an HFS+ boot SSD for now, however, on my production Mac Pro (4,1->5,1 2x X5690 mod that I ebay'd in January 2017 to replace my aging 3,1 at half the cost of an equivalent performance 6,1)). It's good to see Apple throwing us owners of the cheese-grater Mac Pros these bones while we wait for the "modular" macpro7,1 (hoping all the while it's not another 6,1 (nMP) insanity), though I wish these fixes had been distributed long ago.
Many thanks to all of you in this forum thread for teasing out what Apple's been doing in the firmware updates, and testing for the rest us - old aphorism: "you can always tell who the pioneers are - they're the ones with the arrows in their backs."
The Vega’s VBIOS is locked. Or more precise, signed. Any modification will void it. AFAIK, no point to mod that at this moment.
But we can mod the kext inside MacOS to downvolt the card. I never do that because I don’t have to. Mod VBIOS is much easier for me, also no need to perform the mod again and again after MacOS update.
You should able to find the details in some Hackintosh forums, AFAIK, the kext mod is from there. However, you still need Windows to test your card, and copy the parameter from there to the kext.
For cMP, May be simply perform the Pixlas mod is easier if that’s power related. Of course, if thermal related, then we will need some other work around.
Yep, no way to sign a modded Vega VBIOS and Vega pulls very little power from the slot unlike Polaris, perhaps 20W, so a 210W card may pull 95W over each boost connection, which is barely within the cMP's capability. There are a few Vega cards such as Pulse 56 and Red Dragon 56 that are 180W TDP and they have a 165W TDP secondary VBIOS, so those might be okay without Pixlas mod. However, some people have pointed out there are spikes of higher power consumption. RX 580 must have spikes too though and I'm not sure how they compare.
TBH, the numbers from Sapphire looks very strange to me.
The reference RX580:
185W (Max 1340MHz), suggested 500W PSU, single 8pin, dual slot reference cooler
The PULSE RX580:
<225W (Max 1366MHz), suggested 500W PSU, single 8pin, dual slot double 95mm open fan cooler
The Nitro+ RX580
<250W (Max 1430MHz), suggested 500W PSU. 6+8pin, 2.2 slot double 95mm open fan cooler
The reference Vega 56:
210W (Max 1471MHz), suggseted 650W PSU, 8+8pin, dual slot reference cooler
The PULSE Vega 56:
180W (Max 1512MHz), suggested 750W PSU, 8+8pin, 2.5 slots double 95mm open fan cooler
How come the lowest TDP card need biggest PSU?
Why a 180W PULSE Vega card need 8+8pin, but a 225W PULSE Polaris card only has single 8pin?
Why the lowest TDP card has the biggest cooler?
Why the PULSE Vega 56 can draw ~15% less power than the reference card, but can run faster?
For me, the numbers doesn't match. And a quick search got me these. This card is nowhere near 180W TDP but way higher. But still good to see that its cooler can perform well.
When I run Luxmark in Windows GPU-Z shows the Pulse 56 maxing out at 180W. I don’t know how Tom’s Hardware got the card to consume more (unless they changed stock settings). It’s true the Pulse is faster than reference while consuming less power. Perhaps due to better cooling and higher binned chip. I think Pulse 56 has two 8-pins because it takes so little power from the slot. Dual 6 or single 8 is 150W and it doesn’t pull as much as 30W from the slot so that’s why it needs two 8-pins or at least 8+6 like Red Dragon 56. Sapphire probably also gave it a second 8-pin rather than 6-pin because they know their users like to overclock their cards. The Power Color card is more restrictive in that regard and doesn’t like bios flashes. Pulse is more friendly to mod.
Last edited:
AMD cards only give readings from the core power use, they do not give any power reading's for any other part of the PCB that uses power ie ram
not the best sorce but i watch overclocking videos by AHCO
https://www.youtube.com/channel/UCrwObTfqv8u1KO7Fgk-FXHQ/videos
nvidia gpu's give power readings for all power going in to the card.
(he explains it as for volt/shunt modding it matters)
just for fun the sapphire pulse RX 570/580 PCB video
not the best sorce but i watch overclocking videos by AHCO
https://www.youtube.com/channel/UCrwObTfqv8u1KO7Fgk-FXHQ/videos
nvidia gpu's give power readings for all power going in to the card.
(he explains it as for volt/shunt modding it matters)
just for fun the sapphire pulse RX 570/580 PCB video
I have this card:
https://www.dell.com/en-us/shop/del...t/apd/490-beet/graphic-video-cards#polaris-pd
In my system it is recognized as RX480:
Radeon RX 480:
Chipset Model: Radeon RX 480
Type: GPU
Bus: PCIe
Slot: Slot-1
PCIe Lane Width: x16
VRAM (Dynamic, Max): 8192 MB
Vendor: AMD (0x1002)
Device ID: 0x67df
Revision ID: 0x00c7
Metal: Supported, feature set macOS GPUFamily1 v3
Displays:
SyncMaster:
Resolution: 1280 x 1024 (SXGA - Super eXtended Graphics Array)
UI Looks like: 1280 x 1024 @ 60 Hz
Framebuffer Depth: 24-Bit Color (ARGB8888)
Display Serial Number: HMEY510402
Main Display: Yes
Mirror: Off
Online: Yes
Rotation: Supported
Automatically Adjust Brightness: No
I did not upgrade the firmware yet so it is shown as 2.5 GT/s :
Radeon RX 480:
Name: ATY,AMD,RadeonFramebuffer
Type: Display Controller
Driver Installed: Yes
MSI: Yes
Bus: PCI
Slot: Slot-1
Vendor ID: 0x1002
Device ID: 0x67df
Subsystem Vendor ID: 0x1028
Subsystem ID: 0x1701
Revision ID: 0x00c7
Link Width: x16
Link Speed: 2.5 GT/s
The reason I selected this card is because it only occupies 1 slot and has a 6 pin connector only. it is really a Pollaris 10 card running with clock speeds of 1266 which makes it an overclocked RX480:
https://www.techpowerup.com/vgabios/184327/amd-rx480-8192-160603
Known issues:
There is no temperature recognition in the apple firmware (at least in high sierra) so I have to use MACS FAN CONTROL to bump the fans up so the system does not overheat. My next step is to update to BootROM 138.0.0.0.0 to enable the 5 GT/s speeds and HDMI audio.
The card that you linked to isn’t a single slot card, it’s a double. Electrically it connects to 1 PCIe Slot, but physically it occupies a 2 Slot space.
Perhaps you linked a different card than what you intended.
Last edited:
The link to me to an RX 580 for $745.49
Go to the bottom of the post and click that link.
My mistake. it is a 2 slot. Long story short I had RX580 Sapphire, but it occupied 2.5 slots (effectively 3) and it had 6 and 6+2 power connectors so I returned it and replaced it with this DELL AMD 480RX on steroids (branded by DELL as RX580 because it "falls within the OEM specifications for an RX580 card" as per dell own words. Funny how they can go away with such a low level marketing scam. It only cost me 250 bucks on craigslist.
Many current RX 580 cards (including the Sapphire PULSE RX 580 8GB) are above the "standard" 2 slot height and easily encroach on the slot above.
See this thread for my fix to continue using Slot #2 when using an RX 580. Removing two backplate screws and thermal pads "fixed" the issue. Hoping NVIDIA releases Mojave compatible Web Drivers and can stick with GTX 1080 FE.
https://forums.macrumors.com/thread...580-8gb-spacer-wedge-for-pcie-slot-2.2134562/
See this thread for my fix to continue using Slot #2 when using an RX 580. Removing two backplate screws and thermal pads "fixed" the issue. Hoping NVIDIA releases Mojave compatible Web Drivers and can stick with GTX 1080 FE.
https://forums.macrumors.com/thread...580-8gb-spacer-wedge-for-pcie-slot-2.2134562/
A card which is using Ellesmere (Polaris 10) and has only 6 pins is definitely not an RX580 with Polaris 20 with 6+8 pin.
RX 580 ref = 185W.
RX 580 factory OC = 225+
PCI-I = 75W
8pin = 150W
6pin = 75W
confusing, re read the first post.
who did you buy from?
dell direct or ebay?
cant see dell doing anything odd
can see a seller on ebay being odd
anyway a RX480 is the same as a RX580 almost nothing changed.
you always have the option to return if your not happy with it (or how much you paid).
there's going to be almost no real speed change and you always have the option to OC it i gess.
who did you buy from?
dell direct or ebay?
cant see dell doing anything odd
can see a seller on ebay being odd
anyway a RX480 is the same as a RX580 almost nothing changed.
you always have the option to return if your not happy with it (or how much you paid).
there's going to be almost no real speed change and you always have the option to OC it i gess.
Last edited:
Read Dell's statement:
" The AMD Radeon RX560/RX570/RX580 graphics adapter design specification lists a performance range for the card. When vendors build cards to match this specification, there are a wide variety of options that are selected in the implementation, resulting in cards with the same Marketing name but different designs and performances that are still within vendor’s specifications.
Due to this, Dell’s implementation of the AMD Radeon RX560/RX570/RX580 graphics adapter may benchmark slower that some other card models that may land higher in the performance chart range. Dell implementation and performances are within the range specified by AMD for these cards models - the design, power consumption range, memory size range and chip choice are all factors in the card performance.
The following AMD site shows the spec range for the OEM version of the card"
https://www.dell.com/support/articl...-a-amd-radeon-rx460-rx470-rx480-card-?lang=en
As I said already I bought this card locally from craigslist. As a matter of fact the card is recognized in GPU-Z as a RX580 (Windows), but Apple lists it as RX480 in high sierra. My guess is that DELL tweaked the BIOS of the RX480 and overclocked it to fall within the OEM specs. It was designed for Windows so it would be hard to find out the difference if you are not comparing it with a real Polaris 20 GPU.
I prefer the 2 slot card, because anything I put on the top of a 2,5 slot card will block the fan/s and airflow plus the exhaust heat will go straight to the next card on the PCIE slot. That is why I don't need a real RX580, which occupies 2,5 slots. To properly power lets say a card with an 8+6 pin connector you have to draw equal power from both 6 pin connectors on your mac and distribute it properly. I actually bougth a cable splitter from 2 mini 6 pins joined to a single 6 mini pin. You will need another splitter from a single 6 mini pin to the 8+6 pin connector. The 2- 6 mini pin connectors can supply 150W each, but you should not draw unbalanced power from the pins to avoid system shutdowns at peak loads. Of course you can tap straight into the power supply cables, but it requires some rewiring and cable damage.
Just to clarify for anyone who follows this thread later, the Sapphire Pulse RX 580 8GB is powered via single 8-pin, easily achieved via dual mini 6-pin to standard 8-pin cable.
The Sapphire Pulse RX 580 is Apple’s recommended card and please do not confuse newcomers about the power requirements. No 8+6 is required with this card. Your RX 580 variation may be different but that version was not recommended by Apple.
Have posted links several times in the past to the cable I’m using, as have others.
Also posted links about the FIX being used to easily let the 2.25/2.5 slot height work within the Mac Pro tower and still use slot 2 without major issues. It is not ideal and not as great as GTX 1080 FE, but it works and (again) it's Apple’s recommended card. For under $250, officially recommended, easily powered, and works with Mojave, it’s very difficult to recommend anything else.
The Sapphire Pulse RX 580 is Apple’s recommended card and please do not confuse newcomers about the power requirements. No 8+6 is required with this card. Your RX 580 variation may be different but that version was not recommended by Apple.
Have posted links several times in the past to the cable I’m using, as have others.
Also posted links about the FIX being used to easily let the 2.25/2.5 slot height work within the Mac Pro tower and still use slot 2 without major issues. It is not ideal and not as great as GTX 1080 FE, but it works and (again) it's Apple’s recommended card. For under $250, officially recommended, easily powered, and works with Mojave, it’s very difficult to recommend anything else.