Last week we saw what the tiny little Intel Arc A310 was capable of under ideal conditions, today we’re seeing if it can keep up the same pace after we hamstring it.
Welcome to Esk Computers, I’m Scott and today on the desk of Esk, we’re going to be really unfair to this poor little fella. I feel like I’m bullying it.
We’re comparing what’s got to be close to the best possible case with Ryzen 7 9800X3D processor on a B650E-F motherboard with 64GB of 6400MHz DDR5, to this compact, airflow limited case with a Ryzen 5 2400G on a B350i motherboard and 32GB of 2666 MHz DDR4. That’s a massive drop in computing power, but that might not even be the worst of it as far as the A310 is concerned.
While this is an inexpensive piece of kit, manufacturer JGINYUE appear to have backported Resizeable BAR support into their B350i Plus ARGB motherboard, as this wasn’t standard with the B350 chipset. Laudable, however, as the Ryzen 2400G isn’t officially supported for this feature, it’s no help to us here.
The Intel graphics architecture really needs Resizeable BAR access. Why? I was hoping you wouldn’t ask. I’m not going to pretend I understand exactly why Intel designed their cards such that it’s comparatively more important for it to be able to resize the Base Address Register compared to Nvidia or AMD’s offerings, but without it Google tells me the CPU can only access VRAM in tiny 256MB chunks, causing massive frame rate drops and stuttering. And, well, spoiler alert, that’s exactly what this little experiment winds up confirming.
With this setup we’re certainly not just isolating the effects of losing Resizeable BAR support. However there’s already plenty of videos out there covering that, admittedly with the cards it’s more sensible to attempt to game on, but I’m more concerned with something a bit closer to real-world scenario. Arguably.
We do however have some directly comparable results on file for this setup from our first crack at this, testing the Nvidia P1000 and T1000, so that’s almost a level playing field with just the Resizeable BAR situation hampering the A310’s comparative performance. Let’s see what the A310 can manage in a non-ideal scenario, shall we?
As always I’m starting first with the venerable 3DMark Fire Strike, and the A310 is now scoring 5680, dropping 15% from last week’s 6698 points on the 9800X3D system. This means that in the same system it’s down 23% on the T1000’s result of 7333, and 11% better than the P1000’s 5107 points.
Time Spy is up next, with the A310 getting 2661 points, down 16% from its unhindered score of 3165 points. This is down 9% on the T1000’s result of 2936, and 54% better than the P1000’s 1732 points.
This is Superposition at 1080p High Settings, and we are now getting 2875 points, that’s down 11% from last time’s 3216 points. This is down 27% on the T1000’s result of 3960, and 6% better than the P1000’s 2701 points.
The Shadow of the Tomb Raider benchmark at 1080p High sees the A310 now get 24 average fps, down 37% from an already slightly disappointing 38 average fps. This is down 47% on the T1000’s result of 45 fps, and 8% worse than the P1000’s 26 fps.
Counter-Strike 2 with the FPS Heaven Benchmark at 1080p Very High Settings sees us get 47 fps, unusually exactly the same result as the 9800X3D system. This is down 31% on the T1000’s result of 68 fps, and 2% worse than the P1000’s 48 fps.
Dropping to competitive settings, so 1080p Low but with high shadows, the A310 gets 115 fps, down 35% from last week’s 176 average fps. This is down 18% on the T1000’s result of 141 fps, and 7% worse than the P1000’s 123 fps.
I won’t waste our time with ray tracing in Cyberpunk 2077’s benchmark, as It’s struggling enough with 1080p Ultra settings, returning 19 average fps, down 27% from the 26 average fps of last week. That means it is also down 27% on the T1000’s result of 26 fps, albeit 41% better than the P1000’s 13.5 fps.
Dropping down to 720p Low Settings gets us only 39 average fps, down 41% from the last runs 66 average fps. This is down 42% on the T1000’s result of 67 fps, and as it turns out, is exactly the same result that the P1000 manages.
Marvel Rivals at 1080p Native resolution, Medium settings is a demanding test, Resizable BAR or not, and while the 13 fps it’s now getting is only down 7% from the 14 average fps from last week, neither are playable results. Just look at this stuttery mess. This is down 43% on the T1000’s result of 23, and 63% better than the P1000’s 8 fps.
Dropping to Low settings and with Ultra Performance XESS upscaling, we get 31 fps, down 26% from 42 fps. Not only does this look dismal, it’s still a stuttery mess. Not an ideal card for this game even at the best of times.
Sticking with Unreal Engine 5, this is the Black Myth Wukong Benchmark Tool, where at 1080p High, with no RT or FSR, we get 7 fps, 22% down from an already poor 9 fps. This is down 46% on the T1000’s result of 13 fps, and 17% better than the P1000’s 6 fps.
Forza Horizon 5’s benchmark at 1080p Ultra settings gave us 12 fps, down 56% from last week’s 27 average fps, and is another of the very stuttery brigade. Dropping to Medium settings gives us 29 fps, that’s down 52% from the 60 average fps we saw last week. This is down 46% on the T1000’s result of 54, and 22% worse than the P1000’s 37 fps.
Last week we put the A310 through the ringer with some newer games paired with the 9800X3D. The results were at best mixed, so although there’s not a huge amount of value in deliberately making it worse with a much slower CPU and a hobbled graphics card, for science, here’s what we get.
Doom: The Dark Ages sitting at 3fps in the menus is surely not a good sign. By dropping all the way to Low settings and throwing Ultra Performance XESS upscaling at it, we get an 62% lower average framerate of 9 fps, with lows of 1% lows of 4 and 0.1% lows of 3 fps. That’s down from 24 average fps, with 1% lows of 18 and 0.1% lows of 16 fps.
Not that it’s new of course, but for completeness this is 2016’s Doom at its recommended settings, which are mostly Medium, where it gets a 52% lower 23 average fps, with 1% lows of 3, and 0.1% lows of 2 fps. This is down from 48 average fps, with 1% lows of 34 fps and 0.1% lows of 8 fps.
This is Silent Hill f at 1080p Low settings, where it gets its average framerate cut in half, 18 average fps, with 1% lows of 6, and 0.1% lows of 5 fps. Last week it got a comparatively miles better 36 average fps, with 1% lows of 22 and 0.1% lows of 10 fps.
Death Stranding 2 at 1080p Low settings, with ultra performance XeSS enabled also drops well below the almost playable results of last week, with average framerates down 34% to 23 average fps, with 1% lows of 14 fps and 0.1% lows of 10 fps. This is down from 35 average fps, with 1% and 0.1% lows of 25 fps.
Resident Evil Requiem here at 1080p Lowest settings with upscaling off drops 36% with 31 average fps, with 1% lows of 15 and 0.1% lows of 14 fps. That’s down from a decent 49 average fps, with 1% lows of 36 and 0.1% lows of 26 fps.
I’ve saved the worst for last, with Forza Horizon 6’s benchmark at 1080p Low, with no raytracing or upscaling where we now get a miserable 15 fps, down a whopping 72% from last week’s 53 fps result, and is stuttery enough that I don’t see the need to play it properly to declare it as not workable. I’ve also skipped retesting Alan Wake 2 as it barely functioned with all advantages given to it. It’s not going to be playable on this setup.
Well, while that’s more or less what we expected would happen, it’s nonetheless a pretty startling result and pretty shocking seeing it in person. And let’s be honest, while beautiful nerds like you and I knew how big an issue the lack of Resizable BAR support would be, how many normies are going to know what that means?
I suppose those people aren’t the most likely candidates to be building their own PC, especially this sort of compact painbox, but the same issue affects all the Intel cards, and given the price of, well, everything, it’s surely not entirely unreasonable to think of some newbie builder trying to make a low budget build getting hold of an early Ryzen chip and, say, a B570, still one of the only fairly priced low end cards kicking about with halfway acceptable VRAM quantities, and wondering why the end result sucks.
Edge cases, maybe, but there’s enough barriers to entry in this hobby without adding additional pratfall opportunities. Anyway, we’ve confirmed my thinking that the A310’s a very bad pairing with the Ryzen 2400G, and hopefully this serves as a dire warning to prospective builders. I think that’s all the joy we can mine from the A310 so this will probably be sent off to eBay, unless there’s anything you lovely lot want to probe further.
If so, then please leave a comment down below, and if you enjoyed this videotronic missive then consider subscribing. Until next time, take care of yourself, and each other.