The graphics card is the hottest and most heavily loaded component in a computer — especially in gaming and workstation machines. That’s exactly why it lands on our bench most often. The good news: a large share of graphics card failures are repairable; you just need to pin down the real cause. Here’s how it looks from the workshop’s side.
Symptoms that tell you the graphics card is to blame
- Artifacts — blocks, dots, colored lines or stray “pixels” across the image, in a game or even on the desktop.
- Crashes in games or rendering — the game drops, black screen then back, or the computer restarts when the card gets pushed hard.
- No image — the computer runs (fans, LEDs), but the monitor stays black.
- Fans at full blast or dead — noise like a hairdryer, or silence followed by overheating.
- Driver crash — “Display driver stopped responding,” or Code 43 in Device Manager.
What fails most often on a graphics card
Overheating — dried-out paste and thermal pads
Number one. Dust clogs the heatsink, the thermal paste on the chip dries out, the pads on the memory harden — and the card starts to choke: it throttles, throws artifacts, shuts down under load. This is where the right material changes everything. On powerful, hot-running cards (RTX 30/40 series, gaming cards) we use Honeywell PTM7950 — a phase-change thermal material that turns liquid under heat and perfectly fills the micro-gaps, then stays put once it cools. It doesn’t dry out like ordinary paste and doesn’t “pump out”, so temperatures stay low for years. For weaker cards, quality paste is enough; for powerful ones, which overheat the most, PTM7950 is the right choice.
Fans — noise, bearing failure, not spinning
The fan bearing wears out (whistling, rattling) or the fan simply stops. If the card loses its cooling, it overheats fast. Fans get replaced — usually a same-day fix.
GPU chip and memory — reballing
When artifacts or “no image” come from the chip itself or poorly soldered memory, we do a reballing (re-soldering the BGA chip). We’ll be honest with you: reballing isn’t always a permanent solution and isn’t always worth it — after diagnostics we tell you realistically whether it pays off or whether it’s smarter to get another card.
Card power delivery (VRM) and connectors
A short circuit, a burned-out VRM or a damaged PCIe power connector — rarer, but it happens, especially after unstable power or the wrong cable.
How we approach the repair
- Diagnostics — we determine whether it’s the graphics card or something else (power supply, motherboard, driver). Free if it can be seen without disassembly.
- Cleaning + thermal solution — heatsink, paste or PTM7950 on the chip, pads on the memory.
- Fan replacement if one has failed.
- Reballing if needed, only after assessing whether it’s worthwhile.
- Stress test — we run the card for 30–60 minutes under full load and watch temperatures and stability before we return it.
When it’s worth repairing, and when to replace
If the card is fine and was just overheating — cleaning and PTM7950 are worth every dinar; you get years more out of it. If the chip is physically damaged on an old, weak card — it’s often smarter to invest in another. We don’t push a repair at any cost; we tell you honestly what’s worth it.
Frequently asked questions
Can artifacts always be fixed?
Not always. If they come from overheating (dried-out paste or pads) — yes, cleaning and fresh thermal paste fix it. If the chip is physically damaged, reballing is an attempt that doesn’t always succeed. We know the exact condition after diagnostics.
How do I know whether it’s the graphics card or something else?
Signs that point to the graphics card: artifacts, crashes specifically in games and rendering, no image while the computer runs. But the same symptoms can come from the power supply or the motherboard — so diagnostics come first, so you don’t replace the wrong thing.
What is PTM7950 and why do you use it specifically on graphics cards?
Honeywell PTM7950 is a phase-change thermal material: under heat it turns liquid and fills every micro-gap between the chip and the heatsink, then stays stable. Unlike paste, it doesn’t dry out and doesn’t “pump out,” so a powerful graphics card stays cool for years. That’s why we use it specifically on hot-running gaming cards.
Do you do reballing on graphics cards?
We do, but with an honest assessment — reballing only makes sense when it’s cost-effective and there’s a real chance of a lasting result. If that chance isn’t there, we’ll tell you right away, so you don’t throw money away.
How much does graphics card repair cost?
It depends on the fault — cleaning and a thermal solution is one thing, reballing quite another. Ballpark prices are in the price list, and we give the exact price after free diagnostics, before the repair.
Graphics card giving you trouble? Bring it in (or the whole computer) to one of our shops, or a courier picks it up and returns it. More about the service itself on the graphics card repair in Belgrade page.