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

NVIDIA GeForce GTX 1650 Specifications and Price Revealed

btarunr

Editor & Senior Moderator
Staff member
Joined
Oct 9, 2007
Messages
47,677 (7.43/day)
Location
Dublin, Ireland
System Name RBMK-1000
Processor AMD Ryzen 7 5700G
Motherboard Gigabyte B550 AORUS Elite V2
Cooling DeepCool Gammax L240 V2
Memory 2x 16GB DDR4-3200
Video Card(s) Galax RTX 4070 Ti EX
Storage Samsung 990 1TB
Display(s) BenQ 1440p 60 Hz 27-inch
Case Corsair Carbide 100R
Audio Device(s) ASUS SupremeFX S1220A
Power Supply Cooler Master MWE Gold 650W
Mouse ASUS ROG Strix Impact
Keyboard Gamdias Hermes E2
Software Windows 11 Pro
NVIDIA is releasing its most affordable graphics card based on the "Turing" architecture, the GeForce GTX 1650, on the 23rd of April, starting at USD $149. There doesn't appear to be a reference-design (the GTX 1660 series lacked one, too), and so this GPU will be a partner-driven launch. Based on NVIDIA's smallest "Turing" silicon, the 12 nm "TU117," the GTX 1650 will pack 896 CUDA cores and will feature 4 GB of GDDR5 memory across a 128-bit wide memory interface.

The GPU is clocked at 1485 MHz with 1665 MHz GPU Boost, and the 8 Gbps memory produces 128 GB/s of memory bandwidth. With a TDP of just 75 Watts, most GTX 1650 cards will lack additional PCIe power inputs, relying entirely on the slot for power. Most entry-level implementations of the GTX 1650 feature very simple aluminium fan-heatsink coolers. VideoCardz compiled a number of leaked pictures of upcoming GTX 1650 graphics cards.



View at TechPowerUp Main Site
 
And with an uneven number of TPC it will not be the fullfledged Chip.
It is rumoured to have the TU117-300 insted of the full activated TU117-400.

Ti will follow most likely.
 
I'm expecting somewhat interesting overclocking results out of it, due to tame frequencies on the core and the memory, much like the Vanilla GTX 1660.
That said, Its really hard for me to see much success for this card around the area of cards like the RX 580, albeit his multiplicatively higher power consumption
 
896 shaders and 1485/1665MHz compared to GTX 1660's 1408 and 1530/1785MHz makes this about 40% slower. Right in the middle of 1050Ti and RX570.

The niche it fills is 75W TDP so price/performance may not end up being the most important characteristic for it.
The performance gap is rather large though, I would expect a 1650Ti at some point as well.

These simple aluminium coolers will be horrible as usual though :(
 
I'd love to see how this fares against RX 570 looking at some of those custom-design card prices.
 
Too expensive but i guess you must pay for name.
 
Y'know, a table showing how these specs compare to the SKU it replaces would be really useful, and having to google the specs of the 1050 just to see how much of an improvement this card is likely to be is ... well, not what I want from reading a news post. I read TPU for the information, after all. Please provide it.
 
Holy balls, this is screaming low-end
 
wack.
no one buy this.
this belongs in entry-level gaming laptops.

Too expensive but i guess you must pay for name.
wrong.

cards from this segment just never have made any sense price/performance wise,be it amd or nvidia. and lower you go in this sub-$150 segment,the worse it gets. I had radeon r7 250 back in the day and while it was cheap it was the worst investment to my PC I have ever made.sold it for 7870xt that cost 3x as much and I was super happy.
 
wack.
no one buy this.
this belongs in entry-level gaming laptops.
And what if you're fine with entry-level gaming laptop performance, but you have a desktop?

The card is fine. It may rise eyebrows on a niche forum like this one, but it'll be a bestseller. All GTX *50 were.
 
Y'know, a table showing how these specs compare to the SKU it replaces would be really useful, and having to google the specs of the 1050 just to see how much of an improvement this card is likely to be is ... well, not what I want from reading a news post. I read TPU for the information, after all. Please provide it.
63a.jpg
 
And what if you're fine with entry-level gaming laptop performance, but you have a desktop?

The card is fine. It may rise eyebrows on a niche forum like this one, but it'll be a bestseller. All GTX *50 were.
you get a $200 one like 1660 or buy the mid-range one from previous gen (pascal/polaris) at lower price.
 
@btarunr
You should add RX 560 here. That's the actual competitor.
RX570 will compete on performance, but is from a different segment.
 
I bet it performs very close the the rx570 @75w... that's actually amazing

It's not going to OC well but it's not an enthusiast card so it don't matter.

The RX 570 has a $129 variant everywhere so this is going to be a competitive against the 1650...
 
If we assume similar clocks to the 1050Ti this could be ... acceptable. A 16.7% increase in CUDA cores, somewhere around 5% faster on top of that from the improved architecture, but only accompanied by a 14% increase in memory bandwidth (though memory compression is likely somewhat improved). I'm guessing we'll see between 15-20% improvements on average. All for ... $10 less than two and a half years ago. That's not much in terms of progress, no.

Seeing this makes me even more happy with the $80-ish (barely) used RX 570 I got a couple of months back :D

I bet it performs very close the the rx570 @75w... that's actually amazing

It's not going to OC well but it's not an enthusiast card do it don't matter.
According to TPU's GPU database, the RX 570 is 50% faster than the 1050 Ti. This is never, ever getting close to that. The specs simply aren't there.

@btarunr
You should add RX 560 here. That's the actual competitor.
RX570 will compete on performance, but is from a different segment.
Depends how you define segment. Price-wise, it's very close, even if it draws more power. There's no denying Nvidia's efficiency and perf/W lead, but perf/$ is just as important.
 
But as people do care about power consumption, this could be an awesome low-power card as there's no power connector. I'd call this as the 750 Ti of 2019.
 
If it can beat rx570, then it can beat 1060 3gb too, and without power connector that would be awesome. Nice progress with every new generation.
 
But as people do care about power consumption, this could be an awesome low-power card as there's no power connector. I'd call this as the 750 Ti of 2019.
What would be the application? Any PC that has that drastic power issues is a SFF for office or HTPC. None of them need the extra processing power that this card adds, so it won't be a good buy there.

If it can beat rx570, then it can beat 1060 3gb too, and without power connector that would be awesome. Nice progress with every new generation.
It won't beat any of those.
 
But as people do care about power consumption, this could be an awesome low-power card as there's no power connector. I'd call this as the 750 Ti of 2019.
The 750 Ti was revolutionary due to its massive improvement in efficiency compared to previous generations. This won't come close to that, unfortunately. It's more like the GTX 950 of 2019 - perfectly fine, but unremarkable compared to its predecessor.
 
What would be the application? Any PC that has that drastic power issues is a SFF for office or HTPC. None of them need the extra processing power that this card adds, so it won't be a good buy there.
You clearly don't spend much time on SFF forums - there are quite a few people trying to squeeze as much power as possible into tiny boxes with PicoPSUs or similar plug-in DC-DC PSUs, which often have low sustained power limits. The most popular is likely the HDPlex 160, which can sustain 160W and 200W peaks - perfectly fine for a slot-powered GPU and a 65W CPU, but nothing more.

Then again, I've shoehorned my RX 570 into an old Optiplex 990 SFF, which has a proprietary 240W PSU, and it runs perfectly fine alongside its (95W TDP, ~80W according to HWMonitor) i5-2400. The highest power draw I've seen at the wall is 265W, which with the PSU's abysmal efficiency rating of "65% mean" translates to less than 200W internally. Too much for the HDPlex 160, but still not much. I could probably blow up that PSU if I ran Prime95 and FurMark, though.




As for performance, I didn't spot that @londiste had the clock speeds listed further up in this thread, which makes the calculation in my previous post a bit off. Updated, but of course purely theoretical:
GTX 1050 Ti = 100
+16,7% shaders = 116,7
+ ~5% arch improvements = 122,5
+ (1485/1665Mhz base/boost clock vs. 1290/1392 for the 1050 Ti - let's go with boost clocks and call that) 19,6% clock speed increase = 146,5%

But then there's the memory bandwidth wrench thrown in the works, which only has a 14% increase, meaning this is far more bottlenecked by memory than the 1050 Ti, even if we're generous and say improved compression boosts that to 25% better effective bandwidth. My completely pulled-out-of-my-a** estimate (given that calculating the effect of a memory bottleneck takes far more than simple math, and will vary between applications) then lands us at about a 30% improvement over the Ti, smack-dab in the middle of it and the RX 570.
 
75 watts with no need for extra plugs will be the selling point for this one..

trog
 
Back
Top