Nvidia's new GeForce GTX 1630 is a cutdown version of the GTX 1650, that's the same GPU that was released way back in April 2019. To put things in true perspective, we're talking about 3-year-old silicon that to put it mildly was pretty pathetic back then, often losing out to then 2-year older Radeon RX 570.
This GeForce GTX 200mm2 die codenamed TU117 is back, but this time with 42% fewer cores (512 cores), 42% fewer texture mapping units, and half as many ROPs at just 16. The core clock frequency has been bumped up by 7% to 1,785 MHz, but we're getting an embarrassingly small 64-bit wide memory bus with 4 GB of GDDR6 memory for a bandwidth of just 96 GB/s, a 25% reduction from the old GTX 1650.
Based on those figures, the GTX 1630 is going to cost like... $50? What we mean to say is, the original GTX 1650 represented poor value at $150 at launch three years ago, and AMD's recently released Radeon RX 6400 kind of sucks at $160, so for sure something as weak as the GTX 1630 will be priced well under $100 in 2022.
Well, apparently no. Rumor is this model will cost at least $150, with many models said to cost between $170 and $200. In fact, EVGA is currently listing their SC Gaming dual fan model for $200, which is complete madness.
We kind of understood the nonsense pricing for bad products like the Radeon RX 6500 XT during the height of the mining boom, but now that the boom is over and graphics card sales have slowed to a crawl, we don't get what Nvidia's playing at here.
Anyway, we'll talk more about pricing soon, for now let's go over the test system specs. We're using our Ryzen 9 5950X GPU test system -- and yes, we know no one is going to pair a budget graphics card with this CPU -- but the point is to test GPU performance and therefore wish to avoid introducing a CPU bottleneck which would skew the data.
For our low-end to entry-level testing we typically use medium quality settings, or settings that make sense for a given title. We've tested games at 1080p and 1440p, though in this instance we'll focus more on the more relevant 1080p results.
Starting with Assassin's Creed Valhalla using the medium quality preset, we see that the GTX 1630 was good for just 32 fps on average, that's a truly miserable result.
For comparison, the old GTX 1050 Ti was 22% faster and AMD's almost useless RX 6400 was 53% faster using PCIe 3.0, or ~70% faster in a PCIe 4.0 system, a truly embarrassing result for Nvidia.
Shadow of the Tomb Raider was released back in 2018, so it's older than even the TU117 silicon the GTX 1630 is based on, so we test this title using the highest quality preset. The old GTX 1650 was good for 47 fps while the RX 6400 managed 53 fps using PCIe 4.0, the GTX 1630 though was good for just 30 fps. A miserable result for this new GPU.
Watch Dogs: Legion has been turned down to medium, but even here the GTX 1630 was only good for 35 fps on average making the old GTX 1050 Ti almost 10% faster, while the RX 6400 using PCIe 3.0 was a whopping 51% faster, and we can't stress enough how uninspiring AMD's offering is at $160. So what does that say about the 1630?
Rainbow Six Siege is the only game we've been able to play at over 60 fps and we're doing it with the ultra quality preset. Performance was actually comparable to the RX 6400, though the GTX 1050 Ti was still 18% faster.
F1 2021 fans can look forward to sub-60 fps performance using the dialed-down 'high' quality preset. Here the GTX 1630 was almost 10% slower than the RX 6400 using PCIe 3.0 and over 30% slower than the budget Radeon GPU when using PCIe 4.0. It's also worth noting that the GTX 1050 Ti was again much faster, this time delivering 24% greater performance.
Horizon Zero Dawn is rather memory intensive and as a result the GTX 1630 falls off a cliff, or more of an anthill really as it never had far to fall to begin with. Shockingly, the GTX 1050 Ti was 31% faster while the RX 6400 using PCIe 3.0 was an insane 66% faster.
The Far Cry 6 performance was weak as you might expect, though it did at least match the GTX 1050 Ti and that meant the RX 6400 was at least 38% faster, or 81% faster when using PCIe 4.0.
Interestingly, the GTX 1630 actually performed "okay" in Doom Eternal, spitting about 57 fps on average. Still a weak result overall, but decent relative to the Radeon competition, which is very weak. The GTX 1630 did fall short of the GTX 1050 Ti and was slower than the ancient technology that is the Radeon RX 570.
Moving on to Resident Evil Village, we tested using the 'balanced' quality preset which is a medium-type option. The GTX 1630 was capable of just 32 fps on average making the GTX 1050 Ti 22% faster, while the RX 6400 was an absurd 88% faster using PCIe 3.0.
Things couldn't improve in Death Stranding, where the GTX 1050 Ti was 20% faster than the GTX 1630, while the RX 6400 was 75% faster. More truly appalling results for Nvidia's new, but not really new, entry-level graphics card.
As we've come to expect, the GTX 1630 is horrible when testing Hitman 3 as it managed just 27 fps on average with the medium quality settings, making the GTX 1050 Ti 52% faster and the GTX 1650 111% faster.
Lastly, we have Cyberpunk 2077 and this time the GTX 1630 is able to match the 1050 Ti, but with just 22 fps on average at 1080p using the medium quality settings, it was completely unplayable.
The GeForce GTX 1630, like the 1650, is a 75W graphics card, meaning it can do away with external power and just pull what it needs from the PCIe x16 slot. However, for reasons unknown Gainward has stuck a 6-pin PCIe power connector on their 'Ghost' model, meaning external power is mandatory for use with this model.
Basically no GTX 1630 should have external power, though like the GTX 1650, it's possible PCIe slot-only models will be even slower, though surely the 1630 is bloody slow enough. Anyway, the GTX 1630 only uses slightly less power than the GTX 1050 Ti, meaning performance per watt is somehow worse.
12 Game Average
Moving on to the 12 game average data, we see just how inadequate the GTX 1630 is for gaming, delivering 37 fps on average, which means the GTX 1050 Ti is typically 16% faster and the RX 6400 almost 40% faster using PCIe 3.0, or a little over 60% faster using PCIe 4.0. The GTX 1650 is also typically 65% faster and the old RX 570 almost 80% faster.
Cost per Frame
Given the data we've just seen, the cost per frame seems a bit pointless as you obviously shouldn't buy the GeForce GTX 1630, but let's take a look anyway and have a quick discussion. It's rare that a new entry-level GPU scores so poorly, but the 1630 is something quite special -- and obviously not in a good way.
Should this new GPU become available for $150, it will be a disaster, representing the worst value we've seen in a long time. This price sees it come in at a cost of just over $4 per frame, an 18% premium over the already useless GTX 1650 and 22% over the RX 6400 using PCIe 3.0, or 43% more expensive than the PCIe 4.0 configuration.
Right now we recommend gamers spend at least $330 on the Radeon RX 6600. Yes, that's just over twice the price, but that's actually a usable GPU that offers a great gaming experience with almost 4x the performance of the GTX 1630. What's scary is the fact that some GTX 1630 models could sell for as much as $200, resulting in a cost per frame of $5.40, a 90+% premium when compared to the RX 6400 using PCIe 4.0.
Working out how much the GTX 1630 needs to cost in order to match the already underwhelming RX 6400 using PCIe 4.0, the answer to that is $105. Then in order to match the RX 6600 it can't cost more than $87. But given you'd need to play most games using low quality settings, the GTX 1630 needs to be even cheaper, no more than $70 seems reasonable to us.
What We Learned
The GeForce GTX 1630 is an utter embarrassment if marketed as a gaming graphics card. We don't see a single redeeming feature to speak of, and unless it goes on sale for less than $100, it should be completed ignored or used creatively for some Nvidia memes.
If you're after a sub-$200 graphics card, you are far better off turning to the second hand market for the time being. Checking out eBay, you'll find GTX 1650 Super cards regularly selling for around $150. That's as cheap as we imagine the GTX 1630 will be available for and the 1650 Super offers about 125% more performance.
Those not wanting to go second hand are better off digging deep and coughing up the cash for the Radeon RX 6600, which as we just showed is significantly better value despite costing a little over $300.
Finally, if you must buy brand new and you want to spend less than $200, get the Radeon RX 6400. It's a significantly better product despite it's crappy nature in its own right. Nvidia has done the seemingly impossible with the GTX 1630, making the horrible Radeon product look somewhat good in comparison.
We can't wait for the DDR4 version of the GTX 1630 that will no doubt get quietly pushed out in a few months. Until that time, this is the worst GPU we've looked at in quite some time.
- Nvidia GeForce GTX 1630 on Amazon
- Nvidia GeForce GTX 1660 Super on Amazon
- AMD Radeon RX 6400 on Amazon
- Nvidia GeForce RTX 3050 on Amazon
- AMD Radeon RX 6500 XT on Amazon
- Nvidia GeForce RTX 3060 on Amazon
- AMD Radeon RX 6600 on Amazon
- Nvidia GeForce RTX 3060 Ti on Amazon
- Nvidia GeForce RTX 3070 on Amazon