- Joined
- May 13, 2023
- Messages
- 13
Hey everyone! Got my Arc A380. This one is the "Asrock Challenger Arc A380".
This review isn't one like you might expect, part of that is the energy I want to spend on it and the other is I want to try and focus on usability. This will be a usability review for a normal person installing the GPU. Some fun tidbits and oddities will be explained as we go along.
First lets take a look at it. All my pictures where done with my phone so sorry in advance. The box is normal as usual, but I would like to note it did not come with a driver disk, or adapter. This will be important later.
** Taking a look **
All pretty standard so frills so far, which is to be expected for like $140 USD.
This is a picture of it next to the RX580 its going to replace in my eGPU enclosure. As you saw in the backside pic the card is tiny.
** Using the card **
To start out, this was installed in my Razer core X eGPU enclosure. Mine isnt the fancy one with lights and I/O its just a USB-C out. This connects to my work laptop, which is part of the restriction I have in regards to tests. The GPU's primary function in this case is to drive multiple displays and use as little power as possible while and hopefully not as much noise. The requirements are pretty anemic and given that the GPU is entry level other than light gaming, handling video and 3d accelerated apps is probably how most of these will spend their lives.
This is where we ran into our first issue. Remember how the card came with no driver disk? Well thats ok, windows can auto install and download one for us.
It didnt. No amount of Windows updates or optional updates picked up the GPU. As far as Windows was concerned it was 100% an unknown device. This is a big problem for non-IGP systems. Mainly because if you go to format your system you are probably going to be SOL.
Moving on I downloaded drivers ahead of time so lets give it a shot.
To start I download the beta on the front page of TPU. In this case it was "30.0.101.3268".
I opened the device manager so we can take a look at it during the install process. You can see my IGP (also an XE) is already present (again its a laptop). The installation page its a normal EULA. You cant continue unless you scroll down.
The screenshot shows the components that will be installed. Pay attention to this its important later.
The install continues as normal, much like any other. The "ARC Ring" spins and stuff so its something to look at, while they shoot you the same 3 "Fun facts" during the install. Nothing to get excited about, but thy shouldn't really need to either.
The good part comes now, and this actually bothers me. You see by clicking "show More" we can see what the installer is doing. Users might to this because of boredom, more experienced users might use this or logs to troubleshoot or track things. The problem is that the list doesn't scroll, actions and tasks extend out of view and it does not auto scroll. This is a nice to have in general but is important in my case because this is the part where the machine locked up. I took this screen shot and saved it before the machine managed to do it. The display didn't flicker, so I am assuming it was at the actual attempt to reset the display adapter. I am just guessing this may have something to do with the fact I have integrated but that's conjecture. I think maybe the calls to restart are the same but who knows.
This of course set off a chain reaction of issues. I re-ran the installer which was successful and attempted a restart but no such luck. The big one was ARC control itself. The monitors were detected, and the driver was correct. However when attempting to open ARC control we would get errors from the failed install (since the machine needed to be force rebooted the first time.)
Here are some fun errors:
Our driver seems fine and other than the CP though the machine is running normally.
The real shit kicker was the re-install. Everything went fine as you might imagine, but remember when I told you the install options were important? Well you guessed it. When a driver is already detected the ARC drivers only give you an option for a clean install. This is a problem because I only wanted the ARC CP. This would be a minor annoyance except that the clean install doesn't fix the issue. I'm guessing because of left over or broken components. Can we download and install it separately??
Nope.
Thankfully the Intel driver tool detected some driver updates. My machine is managed by the corp so others were outdated. The important ones we want to look at are the integrated and ARC specific drivers. Why do we care about this? Well:
It picks up the older version of the driver. This makes sense because its the WHQL. Now assuming and I might be taking a risk here, that most users aren't going to hit clean install. They will install the driver reboot and move on. So Can we go backwards? Will it fix ARC CP?
The answer is no. The machine locked up again and after a force reboot I needed to rip it out via programs and features.
With the drivers re-installed after a programs and features purge we can now access the ARC cp!
So what does it look like? This part is a bummer. While the ARC cp does see that I have an A380. The drivers for my IGP are what uses the version shown. As you can see in the combo GPU-Z and ARC cp screenshot.
Now, I would be curious to know if this is an issue with the internal card like I mentioned before. We may never know. Moving on, I wanted to see if their was a newer driver. Oh snap!
New driver - https://www.intel.com/content/www/u...tel-arc-graphics-windows-dch-driver-beta.html
It is 1.2GB vs the one on TPU at the time of testing which weighed 865mb. With the driver removed from the windows cp. It was time to test the new one. The screens were identical. As were the problems
but new is new!
Now it was time to beat it up a little. Unlike the gunnir card the Asrock card is clocked 450mhz faster for a total of 2450 clock.
To test it, I simply let is sit on furmark for 2 hours. I mean any card should be able to tolerate it, and this will be mote than this card will normally ever be subjected too, in my system or otherwise.
First a quick warm up to test the waters with GPU-Zs built in.
Finally out furmark 2 hour run. These results are both impressive and confusing.
The MAX power draw was 44w which doesn't align with the power limits or the expectations of the power draw from the card itself. The temps were great though for 2 hours of furmark, though im sure that has a lot to do with the wattage draw. A bit is probably offset by the higher clocks.
One of the last issues is shutting down. Generally in the land of eGPU's this means the card and enclosure fans also shut down. This is NOT the case with the A380, atleast Asrocks version. When shutting down the singular fan ramps up and just hums happily along. This seems like an ACPI issue and maybe more vBIOS related as opposed to driver. Its a heavy annoyance to shut down the enclosure after a shutdown of the machine. Though it should be noted, this may be a bug limited only to enclosures.
** Good **
Its low power
Its cheap
It has 6GB of GDDR6
It has a high clockrate
** BAD **
Drivers WHQL, and 2 beta's worth of problems
Power caps/regulation
Odd power state on shutdown
ARC control panel
Finally it was time to grab the BIOS.
I have a flashcat and a CH341. The 341 was on my desk so thats what I used. The chip is a: Winbound 25Q64JVSIQ but the chip profile I used in ASprogrammer was W25Q64BV since the chip on the A380 is just a revision.
The flash was tested as follows:
2x read > compare > zero chip > verify 0 > write > verify write
The card booted right up so the copy was good.The BIOS is in the attachments below.
The software locks on my machine prevent normal testing, of normal software, the same with my circumstances and configuration. If you would like to see something, and understand it is a work laptop and not a direct comparison to some desktop with a 3090ti let me know!
This review isn't one like you might expect, part of that is the energy I want to spend on it and the other is I want to try and focus on usability. This will be a usability review for a normal person installing the GPU. Some fun tidbits and oddities will be explained as we go along.
First lets take a look at it. All my pictures where done with my phone so sorry in advance. The box is normal as usual, but I would like to note it did not come with a driver disk, or adapter. This will be important later.
** Taking a look **
All pretty standard so frills so far, which is to be expected for like $140 USD.
This is a picture of it next to the RX580 its going to replace in my eGPU enclosure. As you saw in the backside pic the card is tiny.
** Using the card **
To start out, this was installed in my Razer core X eGPU enclosure. Mine isnt the fancy one with lights and I/O its just a USB-C out. This connects to my work laptop, which is part of the restriction I have in regards to tests. The GPU's primary function in this case is to drive multiple displays and use as little power as possible while and hopefully not as much noise. The requirements are pretty anemic and given that the GPU is entry level other than light gaming, handling video and 3d accelerated apps is probably how most of these will spend their lives.
This is where we ran into our first issue. Remember how the card came with no driver disk? Well thats ok, windows can auto install and download one for us.
It didnt. No amount of Windows updates or optional updates picked up the GPU. As far as Windows was concerned it was 100% an unknown device. This is a big problem for non-IGP systems. Mainly because if you go to format your system you are probably going to be SOL.
Moving on I downloaded drivers ahead of time so lets give it a shot.
To start I download the beta on the front page of TPU. In this case it was "30.0.101.3268".
I opened the device manager so we can take a look at it during the install process. You can see my IGP (also an XE) is already present (again its a laptop). The installation page its a normal EULA. You cant continue unless you scroll down.
The screenshot shows the components that will be installed. Pay attention to this its important later.
The install continues as normal, much like any other. The "ARC Ring" spins and stuff so its something to look at, while they shoot you the same 3 "Fun facts" during the install. Nothing to get excited about, but thy shouldn't really need to either.
The good part comes now, and this actually bothers me. You see by clicking "show More" we can see what the installer is doing. Users might to this because of boredom, more experienced users might use this or logs to troubleshoot or track things. The problem is that the list doesn't scroll, actions and tasks extend out of view and it does not auto scroll. This is a nice to have in general but is important in my case because this is the part where the machine locked up. I took this screen shot and saved it before the machine managed to do it. The display didn't flicker, so I am assuming it was at the actual attempt to reset the display adapter. I am just guessing this may have something to do with the fact I have integrated but that's conjecture. I think maybe the calls to restart are the same but who knows.
This of course set off a chain reaction of issues. I re-ran the installer which was successful and attempted a restart but no such luck. The big one was ARC control itself. The monitors were detected, and the driver was correct. However when attempting to open ARC control we would get errors from the failed install (since the machine needed to be force rebooted the first time.)
Here are some fun errors:
Our driver seems fine and other than the CP though the machine is running normally.
The real shit kicker was the re-install. Everything went fine as you might imagine, but remember when I told you the install options were important? Well you guessed it. When a driver is already detected the ARC drivers only give you an option for a clean install. This is a problem because I only wanted the ARC CP. This would be a minor annoyance except that the clean install doesn't fix the issue. I'm guessing because of left over or broken components. Can we download and install it separately??
Nope.
Thankfully the Intel driver tool detected some driver updates. My machine is managed by the corp so others were outdated. The important ones we want to look at are the integrated and ARC specific drivers. Why do we care about this? Well:
It picks up the older version of the driver. This makes sense because its the WHQL. Now assuming and I might be taking a risk here, that most users aren't going to hit clean install. They will install the driver reboot and move on. So Can we go backwards? Will it fix ARC CP?
The answer is no. The machine locked up again and after a force reboot I needed to rip it out via programs and features.
With the drivers re-installed after a programs and features purge we can now access the ARC cp!
So what does it look like? This part is a bummer. While the ARC cp does see that I have an A380. The drivers for my IGP are what uses the version shown. As you can see in the combo GPU-Z and ARC cp screenshot.
Now, I would be curious to know if this is an issue with the internal card like I mentioned before. We may never know. Moving on, I wanted to see if their was a newer driver. Oh snap!
New driver - https://www.intel.com/content/www/u...tel-arc-graphics-windows-dch-driver-beta.html
It is 1.2GB vs the one on TPU at the time of testing which weighed 865mb. With the driver removed from the windows cp. It was time to test the new one. The screens were identical. As were the problems
but new is new!Now it was time to beat it up a little. Unlike the gunnir card the Asrock card is clocked 450mhz faster for a total of 2450 clock.
To test it, I simply let is sit on furmark for 2 hours. I mean any card should be able to tolerate it, and this will be mote than this card will normally ever be subjected too, in my system or otherwise.
First a quick warm up to test the waters with GPU-Zs built in.
Finally out furmark 2 hour run. These results are both impressive and confusing.
The MAX power draw was 44w which doesn't align with the power limits or the expectations of the power draw from the card itself. The temps were great though for 2 hours of furmark, though im sure that has a lot to do with the wattage draw. A bit is probably offset by the higher clocks.
One of the last issues is shutting down. Generally in the land of eGPU's this means the card and enclosure fans also shut down. This is NOT the case with the A380, atleast Asrocks version. When shutting down the singular fan ramps up and just hums happily along. This seems like an ACPI issue and maybe more vBIOS related as opposed to driver. Its a heavy annoyance to shut down the enclosure after a shutdown of the machine. Though it should be noted, this may be a bug limited only to enclosures.
** Good **
Its low power
Its cheap
It has 6GB of GDDR6
It has a high clockrate
** BAD **
Drivers WHQL, and 2 beta's worth of problems
Power caps/regulation
Odd power state on shutdown
ARC control panel
Finally it was time to grab the BIOS.
I have a flashcat and a CH341. The 341 was on my desk so thats what I used. The chip is a: Winbound 25Q64JVSIQ but the chip profile I used in ASprogrammer was W25Q64BV since the chip on the A380 is just a revision.
The flash was tested as follows:
2x read > compare > zero chip > verify 0 > write > verify write
The card booted right up so the copy was good.The BIOS is in the attachments below.
The software locks on my machine prevent normal testing, of normal software, the same with my circumstances and configuration. If you would like to see something, and understand it is a work laptop and not a direct comparison to some desktop with a 3090ti let me know!
Last edited: