Was bored so I decided to do some GT 1030 benchmarks, and I'll say I'm kinda impressed

[TwilightRavens]

Before I start I have to say, no, the GT 1030 is NOT my main GPU in my rig, its simply in there to power the second monitor so I can dedicate as much power as I can to my main card, with that said I decided to do some benchmarks on 3dmark Sky Diver as I don't see many for this class of card (because its so low end) and I wanted to test overclock scaling on the GPU by itself without the CPU being a limiting factor.

 

Disclaimer: I omitted the physics score and combined score as the CPU is not the scope of this test and is not relevant, as its not a real world pairing of CPU/GPU it is simply my fastest CPU and was unlikely to bottleneck the card. The Graphics score and overall score remain, the overall score may or may not be directly comparable to other systems running this GPU depending on 3DMark's CPU score weighing. The GT 1030 is running in the second slot below my main GPU (1080 ti) and it is correct in reporting that it's running 4x on the PCI-E interface, its not the 1080 ti causing that nor the motherboard, the card is simply wired as a PCI-E 3.0 4x card and will run as such, that is highly unlikely to affect the score in any meaningful way even if the card was physically capable of running x8 or x16. It says its running PCI-E 1.1 but that's just the power saving thing, during all the test it was running at PCI-E 3.0 speeds.

 

I will probably update with more results like a manual core oc, but for now MSI's OC scanner will have to suffice. I may also put the card under extreme cooling if I can find a working shunt mod.

 

 

3dMark Skydiver GT 1030 Benchmarks

 

Test System:
CPU: Ryzen 9 3900X (PBO enabled, unlimited power limit)
CPU Cooler: Corsair h100i v2 (with 4x Noctua Redux 1700RPM fans)
GPU: MSI GT 1030 2G LP OC (GDDR5) (Repaste with Thermal Grizzly Cryonaut and stolen thermal pads from my MSI R9 380 Gaming 4G for the VRAM)
GeForce driver version: 445.87
Mobo: AsRock X570 Taichi
RAM: G.Skill TridentZ Neo Samsung B-die Dual Rank kit (3800MHz 16-16-16-16-36 1T 1.38v) fclk, uclk, and ram clock are synced 1:1:1
PSU: EVGA Supernova 750W PQ Platinum rated Semi-modular
OS: Windows 10 Pro 1909 x64

 

Test Results:

Stock Core and Memory:
Graphics Test 1: 55.15 fps
Graphics Test 2: 53.00 fps
Graphics Score: 11,837 pts
Overall Score: 13,066 pts

 

Memory +800MHz:
Graphics Test 1: 56.89 fps
Graphics Test 2: 56.11 fps
Graphics Score: 12,372 pts
Overall Score: 13,621 pts

 

Memory +900MHz:
Graphics Test 1: 57.26 fps
Graphics Test 2: 56.45 fps
Graphics Score: 12,450 pts
Overall Score: 13,702 pts

 

Memory +950MHz:
Graphics Test 1: N/A (driver crash)
Graphics Test 2: N/A (driver crash)
Graphics Score: N/A (driver crash)
Overall Score: N/A (driver crash)

 

Memory +1000MHz:
Graphics Test 1: N/A (system lockup)
Graphics Test 2: N/A (system lockup)
Graphics Score: N/A (system lockup)
Overall Score: N/A (system lockup)

 

Auto OC scanner OC w/ stock memory:
Graphics Test 1: 55.86 fps
Graphics Test 2: 53.13 fps
Graphics Score: 11,927 pts
Overall Score: 13,153 pts

 

Auto OC w/ +800MHz Memory:
Graphics Test 1: 57.92 fps
Graphics Test 2: 56.28 fps
Graphics Score: 12,502 pts
Overall Score: 13,745 pts

 

Auto OC w/ +900MHz Memory:
Graphics Test 1: 58.26 fps
Graphics Test 2: 56.55 fps
Graphics Score: 12,568 pts
Overall Score: 13,817 pts

 

Closing thoughts; The GPU is strictly held to the 30W limit imposed by the vbios of the card, if anyone knows if the shunt mod works on this card let me know and which shunt to bridge as i'd be curious to see what I could get if the card wasn't consistently bouncing off the power limit. The card does allow a max voltage of 1.1120v which is different from other Pascal cards which are held to 1.093v strictly.

 

Also For fun I ran Firestrike Ultra with the max overclock, I didn't expect much so there before anyone asks lol. (see next picture)

 

I highly doubt G-sync is doing much limiting for that card 3dMark, but good looking out.

[KGGaming]

@TwilightRavens I am pretty impressed at that as well but I mean you have a pretty buff cpu with that

[TwilightRavens]

Just now, KGGaming said:

@TwilightRavens I am pretty impressed at that as well but I mean you have a pretty buff cpu with that

Yeah I was trying to eliminate a CPU bottleneck and focus just on the GPU as the limiting factor. The results would probably be a bit different had I thrown it in my old Core2 Extreme rig (which would be a more realistic scenario for a card like that).

[Haro]

3 minutes ago, KGGaming said:

that as well but I mean you have a pretty buff cpu with that

GPU benchmarks hardly relies on the cpu

That doesn't matter 

A good cpu is to remove any Bottleneck from the equation

[KGGaming]

Just now, TwilightRavens said:

Yeah I was trying to eliminate a CPU bottleneck and focus just on the GPU as the limiting factor. The results would probably be a bit different had I thrown it in my old Core2 Extreme rig (which would be a more realistic scenario for a card like that).

I would be pretty interest in that test with the Core 2, but at the moment Im amazed how a like 5 year old card can do that

[WereCatf]

7 minutes ago, TwilightRavens said:

if anyone knows if the shunt mod works on this card let me know and which shunt to bridge as i'd be curious to see what I could get if the card wasn't consistently bouncing off the power limit

It's hard to say much from such low-quality pictures, especially since one can't actually read any of the markings, but the only resistor that looks like it might be a shunt is on the bottom-left, next to U1. Could always check its resistance with a good-quality multimeter and then the voltage across it when the card is running to check if it is the shunt or not.

[TwilightRavens]

Just now, WereCatf said:

It's hard to say much from such low-quality pictures, especially since one can't actually read any of the markings, but the only resistor that looks like it might be a shunt is on the bottom-left, next to U1. Could always check its resistance with a good-quality multimeter and then the voltage across it when the card is running to check if it is the shunt or not.

Yeah sorry, the camera on my phone isn't the greatest.

[aaradorn]

Even budget GPU's are pretty strong these days. It's quite impressive.

[TwilightRavens]

1 hour ago, KGGaming said:

I would be pretty interest in that test with the Core 2, but at the moment Im amazed how a like 5 year old card can do that

I will actually do that once I find a working EVGA or XFX 790i Ultra SLI board for a non stupid price (i.e. $100 or less and NOT from China) to replace my current fried one, the LGA 771 Xeon I had in there could hang with a stock 2500K with its overclock (that's probably why its fried now that I think about it...)