• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

NVIDIA GeForce GTX 1650 Memory Size Revealed

Let's hope it doesn't. Why pay more for something a particular client group will not use?
I have 750MB GPU memory allocated at the moment (500MB by Firefox for whatever reason). Seriously, 4 GB?

1650 got expensive. I wonder if they'll give us something below in this generation (1630?)
Are you seriously defending 2GB gaming cards... in 2019? Anything with GTX or RTX before the name should be 4GB absolute minimum. GTX 1050 2GB was already being limited by its vram when it launched a couple years ago. 2GB on a card which is going to perform like 1060 6GB in games (if not VRAM bound) is just ridiculous. The GTX 1650 is not aimed at people who want a display output and browse the web. Go get a 1030 for that.
 
No. I simply want to put a graphics card into an office PC (with a CPU that doesn't have IGP).
And since this PC won't be used for gaming or GPGPU, why would I have to pay for 4GB? I might never use more than 1GB.

Why deadset on this gen in the first place then? Plenty of options from last gen and even more from generations prior to that.
 
If you need a good GPU for real work, a 1650 is too little for it. If you need a GPU for desktop use, a 1650 is too much for it.
The market for the 1650 is games at low resolutions, less than that IGP and low end cards, more than that, Quadro or Titan, with their certified drivers.
No offense, but are you aware that some applications utilize GPUs? The list includes a few mainstream Adobe programs (great impact) and MS Office (still somehow problematic but should be fixed in 2019).
I'm not talking about explicit GPGPU computation done by the user.
Are you seriously defending 2GB gaming cards... in 2019?
No. I'm defending 2GB cards. Sans "gaming".
The GTX 1650 is not aimed at people who want a display output and browse the web. Go get a 1030 for that.
That's exactly what the argument is about! We haven't seen a Turing successor to 1030 yet, despite obvious efficiency gains that would be welcome. I'm expecting near 1050 performance with sensible passive cooling.

I don't understand why some people are so much against products aimed at business and casual PC users. It's not harming your 4K60fps monsters.

Why deadset on this gen in the first place then? Plenty of options from last gen and even more from generations prior to that.
Because Turing is a lot faster than Pascal - not to mention older GPUs. As long as cheap cards were only good for video signal, a GT710 could have been the last GPU a business or casual user needed in his life. But since GPU acceleration becomes more and more popular, this sector can also benefit from new tech. :)
 
Last edited:
No offense, but are you aware that some applications utilize GPUs? The list includes a few mainstream Adobe programs (great impact) and MS Office (still somehow problematic but should be fixed in 2019).
I'm not talking about explicit GPGPU computation done by the user.

No. I'm defending 2GB cards. Sans "gaming".

That's exactly what the argument is about! We haven't seen a Turing successor to 1030 yet, despite obvious efficiency gains that would be welcome. I'm expecting near 1050 performance with sensible passive cooling.

I don't understand why some people are so much against products aimed at business and casual PC users. It's not harming your 4K60fps monsters.


Because Turing is a lot faster than Pascal - not to mention older GPUs. As long as cheap cards were only good for video signal, a GT710 could have been the last GPU a business or casual user needed in his life. But since GPU acceleration becomes more and more popular, this sector can also benefit from new tech. :)
If you only need GPU acceleration, performance is not a problem, anything over 128 shaders has more than enough power. And in that case, power consumption is the priority. You need a $100 GT1030 successor, the 1650, as I've been saying, is too much too expensive for that job.
 
Let's hope it doesn't. Why pay more for something a particular client group will not use?
I have 750MB GPU memory allocated at the moment (500MB by Firefox for whatever reason). Seriously, 4 GB?

1650 got expensive. I wonder if they'll give us something below in this generation (1630?)

If you only need 2GB of VRAM, you simply don't need a dedicated GPU at all.

A 1030 is the equivalent to a 2200G. Why spend $85(GDDR5 version) for a dedicated GPU when you can spend $95 for an R3 2200G?

The 1630 would only make sense if it was as fast as a 1050 or higher, and only at 4GB. Everything below the performance of the 1050 has been obsoleted by the R3 2200G.
 
I have R7 240's in my T3500's. The are 25W cheaper to run than the GT 710 I initially bought.
 
If you only need 2GB of VRAM, you simply don't need a dedicated GPU at all.

A 1030 is the equivalent to a 2200G. Why spend $85(GDDR5 version) for a dedicated GPU when you can spend $95 for an R3 2200G?
Because I want a more powerful CPU and in could mean no IGP version is available. :)
The 1630 would only make sense if it was as fast as a 1050 or higher, and only at 4GB. Everything below the performance of the 1050 has been obsoleted by the R3 2200G.
I think you're slightly too focused on gaming here. For games you need a good compromise between CPU and GPU, because one will limit the other. And yes, AMD APUs usually offer very good CPU-GPU relation (because they're designed to do so).
It's very different outside of gaming, when you're usually focusing on CPU performance (GPU-focused machines exists but are rare).
The fastest APU at the moment is 2400G. It's not bad, obviously, but at the moment it lags behind mid-range i5 (the workhorse of office PCs).
We'll see what the Zen2 future brings, but at the moment it's pretty one-sided. Intel is able to pair an IGP with 6 or 8 cores and AMD is limited to 4 (slower ones as well).
 
I wonder what kind of CPU driven work you're doing where you would pick an i5 over a 2400G for a basic office machine. If you need that much CPU grunt, you're moving into workstation territory...
 
Because I want a more powerful CPU and in could mean no IGP version is available. :)

I think you're slightly too focused on gaming here. For games you need a good compromise between CPU and GPU, because one will limit the other. And yes, AMD APUs usually offer very good CPU-GPU relation (because they're designed to do so).
It's very different outside of gaming, when you're usually focusing on CPU performance (GPU-focused machines exists but are rare).
The fastest APU at the moment is 2400G. It's not bad, obviously, but at the moment it lags behind mid-range i5 (the workhorse of office PCs).
We'll see what the Zen2 future brings, but at the moment it's pretty one-sided. Intel is able to pair an IGP with 6 or 8 cores and AMD is limited to 4 (slower ones as well).

Like hat's post above, I'm struggling to see what office PC you're dreaming up that needs more than a 4/4 or 4/8 CPU where an IGP wouldn't be sufficient. Office PC's(excel, word, etc.) rarely need more grunt than 2/4, let alone 4/4 or 4/8.

If you're thinking about a workstation PC that needs a bunch of CPU power but doesn't have an IGP(x299/x399 platforms), those are better off with a professional card with guaranteed support & better general overall stability, a AMD Radeon Pro WX 2100 can be had for $99-115 in that case.
 
The market for the 1650 is games at low resolutions, less than that IGP and low end cards, more than that, Quadro or Titan, with their certified drivers.
Wat
 
Bad use of punctuation. I mean the 1650 is for low budget gamers. Anything less than that and you should use an IGP or a low end card for desktop work.
English is not my native language and I was on the phone, stupid autocorrect in spanish doesn't help.
 
  • Like
Reactions: hat
I wonder what kind of CPU driven work you're doing where you would pick an i5 over a 2400G for a basic office machine. If you need that much CPU grunt, you're moving into workstation territory...
I'm doing pretty standard data analysis and machine learning. And once in a while something more advanced if I'm lucky enough.
Tools typical as well: R, Python, Excel, Tableau.

Just to make it clear: you can choke any CPU available today with a complex (but still sensible enough) spreadsheet file. People all over the world are spending a big chunk of their working time waiting for Excel to recalculate. You replace a humble 4-core i5 with a 20-core Xeon and 3 minutes become 30 seconds. Still not exactly comfortable, but at least you don't fall asleep.
 
Back
Top