Synthetics

Moving on to synthetic performance testing, there shouldn’t be any surprises here. This is the 4th TU116 card we’ve looked at – albeit the first with just a 128-bit memory bus – so its behavior is generally well understood.

Synthetic: TessMark - Image Set 4 - 64x Tessellation

Synthetic: Beyond3D Suite - Pixel Fillrate

Synthetic: Beyond3D Suite - Integer Texture Fillrate (INT8)

Synthetic: Beyond3D Suite - Floating Point Texture Fillrate (FP32)

Synthetic: Beyond3D Suite - INT8 Buffer Compression

Synthetic: Beyond3D Suite - FP32 Buffer Compression

With synthetic performance, we see the GTX 1650 Super benefitting significantly with tessellation throughput, most likely due to the inclusion of more GPCs to house the greater SM count. This still leaves it behind the GTX 1660 cards, but relatively close overall. Meanwhile we see just a slight uptick in pixel throughput, reflecting the fact that while the card has a good deal more memory bandwidth than the regular GTX 1650, for this simple (and generally well compressible) test, it doesn’t add much.

Compute Power, Temperature, & Noise
POST A COMMENT

67 Comments

View All Comments

  • StevoLincolnite - Sunday, December 22, 2019 - link

    But if you only paid $100 and it only lasts for a couple years, it's still worth it for that tier of performance, no? Reply
  • Yojimbo - Monday, December 23, 2019 - link

    We don't have any data on this, which is why I would avoid a used mining card. We'll never get any data on this, either. The thing is, although one can't tell a well-cared-for gaming card from one not well-cared for, over the years a general knowledge of the expectation of a used part has been built up. In the case of mining it is a big unknown in my view. You don't really know which cards are mining cards and which are gaming, so any card that has been popular with miners is suspect, in my view, unless you know who you are buying from. Reply
  • eastcoast_pete - Sunday, December 22, 2019 - link

    Which poses this question: Is there a program ("app") that can run a health check on a card? In addition to any "custom BIOS" , I would also be concerned about simple aging with intense, ongoing use. When manufacturers bin chips and assign them to target speeds, they supposedly do so also based on life expectancy, at least for CPUs. So, is there a way to test how much life the GPU and RAM of a card have left in them? Reply
  • flyingpants265 - Sunday, December 22, 2019 - link

    The elephant in the room for the RX580 8GB, and AMD videocards in general, is the almost 200W power draw on a "1080p card", whereas the 1650 uses 75W. It may really suck for the price, but it uses less than half the power. Obviously the RX570 is a great choice as well.

    Then there's reliability. I've seen statistics from Puget systems and some big online retailer and AMD had some obscenely high failure rates. AMD is a much smaller company, they might have less oversight, and heat causes a lot of damage to complex electronics. Not exactly reliable info but I wouldn't really be surprised if it were somewhat accurate. I believe all consumer products are cheaply made, so I'd rather go with the lower-power, lower-heat, larger company. Too bad I don't have any hard data to back that up.

    Not really interested in anecdotal evidence either.
    Reply
  • Spunjji - Monday, December 23, 2019 - link

    "Too bad I don't have any hard data to back that up.
    Not really interested in anecdotal evidence either."

    Next time start with that pitch, so folks can ignore the self-confessedly uninformed speculative rambling that follows.

    If the power draw is a bother on the RX580, a little undervolting will go a very long way without noticeably affecting frame rates at 1080p. It'll also help with longevity. Regardless, none of this is particularly crucial when you're saving ~$50 and getting a faster card with more VRAM.
    Reply
  • Spunjji - Monday, December 23, 2019 - link

    RX 580 is still good for 1440p too, if you're not obsessed with hitting "Max" on every setting just because it's there. Reply
  • khanikun - Monday, December 30, 2019 - link

    Reminds me of back in the day, when I moved to ATI for a very short while. 9700 Pro. Started overheating after a year, then broke. 9600 XT as temporary card. Started overheating in less than a year. 9800 XT. Started overheating in a couple months. Went back to Nvidia and haven't had a reason to look at ATI/AMD cards since. Reply
  • Qasar - Wednesday, January 1, 2020 - link

    i have some of those cards from then, 9800pro, 9600pro, used them not to long ago to see if they still work.. and they still do.. hehehehe Reply
  • WetKneeHouston - Monday, January 20, 2020 - link

    I think it makes sense to think that heat and power would lead to reliability issues. That's why I went with the 1650 Super. It's still too powerful of a card for the low tier (I can't notice the difference with higher settings, I suspect it's a scam lol) 1080p gaming I do, so I probably should have gotten the regular 1650, but they're basically the same price, Reply
  • Yojimbo - Friday, December 20, 2019 - link

    Yeah maybe if it cost $50 it would be worth it for running dosbox. Reply

Log in

Don't have an account? Sign up now