GeForce GTX 1650 Review: Turing at $150

[grumblguts]

For power saved vs the rx 570 some 60 odd watts and being slower.
You would have to game 24,7 for years none stop to break even.
What a dumb card they actually think peoples hard drives dont have a obsolete cd drive molex cable within the system. I honestly do not understand the point in making a gpu within external power.

 

[Veiran]

So this article proves that the RX580 is and was the better option to the GTX 1060, yet Steam shows that the 1060 outsold nearly 15:1

Why are consumers such *****s.

You can't blame them. For over a year the 580 was much more expensive and with low stocks.

That's a good point! The crypto mining craze very likely took a big chunk out of the potential market share that AMD would've had had the prices remained at MSRP.

I'd completely forgotten about that. And here I was wondering why Nvidia cards are so highly represented on the Steam charts...

 

[Lew Zealand]

The inclusion of a puny GTX 1050 in these results begs for inclusion of even older... yet more powerful... cards as a reference point.

I have an old GTX670... which is faster than a 1050 Ti... and have been looking into "upgrading" for over a year, but I have yet to see a new card that justifies the expense based on the minimal performance improvement.

If a slow 1050 deserves inclusion in these charts, then maybe we should include an old 690 in the tests as well if for no other reason than to show just how little card speed has improved in the past seven/nine years.

A GTX x70 series card shouldn't be compared to an GTX x50 series other than for informational purposes, they are different product segments. Ideally a 670 would be compared to a current x70 series card but with the Nvidia price increases the closest match is probably the 1660Ti. Which I'm sure will blow the 670 away. Same goes for the 2080 (or Ti) vs the 690.

 

[LogiGaming]

You sir, are drinking the coolaid. The best card money can buy, in term of value, is a RX 570, a RX 580 or a Vega 56 below 300$.

In term of performance, it is the 2080 TI or the Titan...

You should remove your Nvidia googles.

And as power goes... Performance is always more important than Power. Nobody buy a card for is power consumption for gaming. We are talking about 5-10$ of electricity per year of difference. It is a false debate. I always base my purchase on price/performance, never in term of power. I take a Vega 64 at 400$ everyday over a 550-600$ 2070.

Imagine thinking that US is the only country in the world.

GTX 1660 overclocked is the best value for the money. GamerNexus concluded the same, sorry. You can call it what you want.

I would never buy a Vega 64. That card is rubbish. And yes power matters too, unless you ignore it because it goes against your argument or favourite brand. Talking about drinkimg the coolaid....

 

[144hzGamer]

GTX 1660 still the best card money can buy. Even if this website states otherwise. I can find them for 230€ here in Europe with Black Ops 4 Bundled. That card is the best overclocker since the gtx 460 days (if someone remembers), and even when compared to a 1660ti max overclocked, it is behind by 5%.

RX 570 is cheap and great performance for the price, but it uses more power wich can be a concern on a lot of people systems, on this price range. A lot of people buying cheap cards can´t upgrade their PSU or are using HTPCs.

GTX 1650 should have been a 100€-120€ GPU at max. At this price is DOA. But I would make the effort to save up for a GTX 1660, by far. No other card can beat its value, if you use MSI Afterburner and drag a slider to the right.

You sir, are drinking the coolaid. The best card money can buy, in term of value, is a RX 570, a RX 580 or a Vega 56 below 300$.

In term of performance, it is the 2080 TI or the Titan...

You should remove your Nvidia googles.

And as power goes... Performance is always more important than Power. Nobody buy a card for is power consumption for gaming. We are talking about 5-10$ of electricity per year of difference. It is a false debate. I always base my purchase on price/performance, never in term of power. I take a Vega 64 at 400$ everyday over a 550-600$ 2070.

Prices on Europe:

RX570 - 160€ and for the 4gb model
Gtx 1660 - 220€

Easy choice. The latter has 40% more performance at stock and 60% when both overclocked, according to TpU and GN. The latter one uses less wattage as a bonus.

Prices in america are different but you need get out of your own bubble pal.

 

[Stark]

So this article proves that the RX580 is and was the better option to the GTX 1060, yet Steam shows that the 1060 outsold nearly 15:1

Why are consumers such *****s.

At least in my region AMD overpriced themselves in comparison to Nvidia at launch of 480, and even though I was planning to upgrade to it it really irked me off. Hence Nividia 1060 gaming x was it is which was priced lower than 480 reference card.

 

[Theinsanegamer]

Yikes ......
(besides the low power comsuption)
RX 570 is superior option (since can find much cheaper and have 8GB models).....oh no guys lets wait on the (we all know will come sooner than later NVIDIA segmenting/saturating the market) 1024~1280 core GTX 1650 TI* guys...LOL

The 1650's lack of external power is its biggest boon. If you have a OEM system without a 6 pin connector, the 1650 is going to be a LOT faster then a 570, considering a 570 wont run. Same goes for low profile systems, where 6 pin connectors are not an option, and unless you have a system that can run, and managed to find, one of the ultra rare low profile 570s, the 1650 (when it gets a low profile model) will be the fastest low profile model available.

 

[QuantumPhysics]

You should go no lower than an RTX2060.

 

[Evernessince]

The 1650's lack of external power is its biggest boon. If you have a OEM system without a 6 pin connector, the 1650 is going to be a LOT faster then a 570, considering a 570 wont run. Same goes for low profile systems, where 6 pin connectors are not an option, and unless you have a system that can run, and managed to find, one of the ultra rare low profile 570s, the 1650 (when it gets a low profile model) will be the fastest low profile model available.

You may want to get a different card if you want a card that doesn't have a PCIe connector.

Steve tested the 1650s without a 6 pin connector and the performance is a dip below what the 1650s in this review scored, which do have that connector. I t may be wiser to simply buy last gen.

I don't get the comparison to low profile cards. There are no low profile 1650s. If you need one, you have to look elsewhere.

Imagine thinking that US is the only country in the world.

GTX 1660 overclocked is the best value for the money. GamerNexus concluded the same, sorry. You can call it what you want.

I would never buy a Vega 64. That card is rubbish. And yes power matters too, unless you ignore it because it goes against your argument or favourite brand. Talking about drinkimg the coolaid....

This is a factually incorrect statement

https://www.techspot.com/review/1811-nvidia-geforce-gtx-1060/

The RX 570 is in fact still the best value. Drawing a conclusion based on a single card's or even a few card's overclocking results would be ill advised given the sample size doesn't give them the test resolution to declare the data accurately reflects the whole picture. Stock results are by far the best indicator of performance as you can always guarantee to get that amount.

 

[LogiGaming]

You may want to get a different card if you want a card that doesn't have a PCIe connector.

Steve tested the 1650s without a 6 pin connector and the performance is a dip below what the 1650s in this review scored, which do have that connector. I t may be wiser to simply buy last gen.

I don't get the comparison to low profile cards. There are no low profile 1650s. If you need one, you have to look elsewhere.



This is a factually incorrect statement

https://www.techspot.com/review/1811-nvidia-geforce-gtx-1060/

The RX 570 is in fact still the best value. Drawing a conclusion based on a single card's or even a few card's overclocking results would be ill advised given the sample size doesn't give them the test resolution to declare the data accurately reflects the whole picture. Stock results are by far the best indicator of performance as you can always guarantee to get that amount.

Pretty much every review we can see of the GTX 1660 shows that it hits at least 2000mhz, some models hit 2100mhz and yes, that value is not assured, but 2000mhz is assured. And with 2k on the core, it is basically a gtx 1660 ti minus 2%. Watch GamerNexus review please.

So funny that we are all tech saviors and always looking for the best deals, but using a MSI Afterburner slider to get more out of our money is not "usual". Right...

If AMD had a 230€ card that could have same performance as a 1070ti by overclocking everyone would be bragging about it.

 

[Evernessince]

Pretty much every review we can see of the GTX 1660 shows that it hits at least 2000mhz, some models hit 2100mhz and yes, that value is not assured, but 2000mhz is assured. And with 2k on the core, it is basically a gtx 1660 ti minus 2%. Watch GamerNexus review please.

So funny that we are all tech saviors and always looking for the best deals, but using a MSI Afterburner slider to get more out of our money is not "usual". Right...

If AMD had a 230€ card that could have same performance as a 1070ti by overclocking everyone would be bragging about it.

I heard the same "every card hits such and such" with the 1080 Ti and the Ryzen 1700 and both turned out to be false when an actual dive into the data was done. Silicon lottery was a good source of what people were actually getting (which was around 3.7 - 3.8 GHz) and the 1080 Ti which hit 2000 only about 30% of the time despite overclocking proponents claiming it's a guarantee.

I've had people assure me that the 1080 Ti can hit 2000 MHz before until I showed them benchmarks of it running well under that value. Silicon quality varies, I need to see data on actual numbers people are getting first before I made some big statement like every GPU getting an overclock.

 

[LogiGaming]

Lol my 1080ti actually hits 2000mhz without any problem, I don´t know about a single 1080ti that didn´t hit 2000mhz on the core, unless you talking about the MSI Armor wich can´t even sustain good temps.

Also I am telling you GTX 1660 usual overclock is 2100mhz, so I put 2000mhz on the low side. even the cheapest models all got 2050mhz on Techpowerup reviews.

I understand not everyone overclocks, but overclocking a GPU is very easy and GTX 1660 once overclocked is a great deal because this card is a beast when overclocked.

 

[RageofThunder3]

Lol my 1080ti actually hits 2000mhz without any problem, I don´t know about a single 1080ti that didn´t hit 2000mhz on the core, unless you talking about the MSI Armor wich can´t even sustain good temps.

Also I am telling you GTX 1660 usual overclock is 2100mhz, so I put 2000mhz on the low side. even the cheapest models all got 2050mhz on Techpowerup reviews.

I understand not everyone overclocks, but overclocking a GPU is very easy and GTX 1660 once overclocked is a great deal because this card is a beast when overclocked.

Based on this logic, a Radeon VII which overclocks like a champ is almost matching the performance of 2080ti and therefore should be the best buy on the high-end market. Give me a break...
Nvidia and AMD know their chips better than you'll ever be.

 

[Evernessince]

Lol my 1080ti actually hits 2000mhz without any problem, I don´t know about a single 1080ti that didn´t hit 2000mhz on the core, unless you talking about the MSI Armor wich can´t even sustain good temps.

Also I am telling you GTX 1660 usual overclock is 2100mhz, so I put 2000mhz on the low side. even the cheapest models all got 2050mhz on Techpowerup reviews.

I understand not everyone overclocks, but overclocking a GPU is very easy and GTX 1660 once overclocked is a great deal because this card is a beast when overclocked.

For one, JokerProductions didn't

1835 at max. That's FAR from 2000 MHz.

And another

And even TechSpot's own 1080 Ti at launch only hit 1785. Skip to 14:51 in the video

Even a factory overclocked high end 1080 Ti failed to reach 2000 MHz

Your comment is the perfect example of why one off overclocking results are worthless in the grand picture. Contrary to your tightly held beliefs, a majority of the reviews I've seen on the 1080 ti that actually displayed GPU frequency never reached 2000 MHz. This especially goes for lower end cards, that can fall up to 200 MHz short of that, which is a significant performance difference.

 

[LogiGaming]

For one, JokerProductions didn't

1835 at max. That's FAR from 2000 MHz.

And another

And even TechSpot's own 1080 Ti at launch only hit 1785. Skip to 14:51 in the video

Even a factory overclocked high end 1080 Ti failed to reach 2000 MHz

Your comment is the perfect example of why one off overclocking results are worthless in the grand picture. Contrary to your tightly held beliefs, a majority of the reviews I've seen on the 1080 ti that actually displayed GPU frequency never reached 2000 MHz. This especially goes for lower end cards, that can fall up to 200 MHz short of that, which is a significant performance difference.

1- Joker videos? No thanks

2- The cards were not manually overclocked

3- The card is a reference card and hits temp limit quick

4- The card was not mannually overclocked.

Your videos are the perfect example of someone trying to dodge and finding videos to base his opinion.

I can do the same:

GTX 1080ti manually overclocked 1:

2: https://www.youtube.com/watch?v=JLOi9po42iA

3: https://www.youtube.com/watch?v=SblOAGuzWFw

4: https://www.youtube.com/watch?v=zzrZC8yHlgU

Based on this logic, a Radeon VII which overclocks like a champ is almost matching the performance of 2080ti and therefore should be the best buy on the high-end market. Give me a break...
Nvidia and AMD know their chips better than you'll ever be.

Except that´s not true? Techpowerup even repasted their Vega VII, did all the possible tricks to decrease temps down by 10º and they still couldn´t manage more than a 8% overclock, wich made it still slower than a GTX 1080ti, nevermind a RTX 2080ti: https://www.techpowerup.com/reviews/AMD/Radeon_VII/33.html

Now let´s look at GTX 1660 overclocked performance according to GamerNexus review.

Almost same performance as a GTX 1660ti max overclocked too, minus 2-3fps. For 220€ + Black Ops 4 it completly obliterates RX590, RX580, RX570 and GTX 1060.

Next arguments please.

 

[144hzGamer]

For one, JokerProductions didn't

1835 at max. That's FAR from 2000 MHz.

And another

And even TechSpot's own 1080 Ti at launch only hit 1785. Skip to 14:51 in the video

Even a factory overclocked high end 1080 Ti failed to reach 2000 MHz

Your comment is the perfect example of why one off overclocking results are worthless in the grand picture. Contrary to your tightly held beliefs, a majority of the reviews I've seen on the 1080 ti that actually displayed GPU frequency never reached 2000 MHz. This especially goes for lower end cards, that can fall up to 200 MHz short of that, which is a significant performance difference.

Are you really showing max gpu clocks on a card that wasnt manually overclocked? You also expect to buy a 9700k and that it clocks automatically at 5,2ghz? Dude... LOL your arguments are getting weird.

Most gtx1660 go up to 2050/2100mhz, so 2000mhz ? That should be piece of cake.

Gtx1660 is a card that has memory bandwidth limitations and comes with pretty low clocks for what it is capable of doing. Once you overclock this card you have a great performance. And yes every GPU overclocks, going by that logic, true. But gtx1660 overclock potential ia above average, thus why it is mentioned.

 

[Evernessince]

1- Joker videos? No thanks

2- The cards were not manually overclocked

3- The card is a reference card and hits temp limit quick

4- The card was not mannually overclocked.

Your videos are the perfect example of someone trying to dodge and finding videos to base his opinion.

I can do the same:

GTX 1080ti manually overclocked 1:

2:

3:

4:

Except that´s not true? Techpowerup even repasted their Vega VII, did all the possible tricks to decrease temps down by 10º and they still couldn´t manage more than a 8% overclock, wich made it still slower than a GTX 1080ti, nevermind a RTX 2080ti: https://www.techpowerup.com/reviews/AMD/Radeon_VII/33.html

Now let´s look at GTX 1660 overclocked performance according to GamerNexus review.

Almost same performance as a GTX 1660ti max overclocked too, minus 2-3fps. For 220€ + Black Ops 4 it completly obliterates RX590, RX580, RX570 and GTX 1060.

Next arguments please.

You linked a Lightning Z, one of the most expensive binned 1080 Tis and the market. That card had a $200 price premium over other 1080 Tis at launch. The fact that I have to explain this means you are either ignorant or trolling.

The video below that doesn't even clearly show frequency

The last video is the only one where they legitimately OC past 2000 but mind you once again on a higher end card.

Thanks for submitting more evidence to my point, you are clearly marking a trend where only the high end 1080 Tis can typically reach over 2000 MHz.

Oh and please submit videos from legitimate reviewers next time. Crappy quality videos of people running a single benchmark isn't a sign of a stable OC. I've had plenty of unstable OCs that worked perfectly fine in firestrike.

 

[LogiGaming]

You linked a Lightning Z, one of the most expensive binned 1080 Tis and the market. That card had a $200 price premium over other 1080 Tis at launch. The fact that I have to explain this means you are either ignorant or trolling.

The video below that doesn't even clearly show frequency

The last video is the only one where they legitimately OC past 2000 but mind you once again on a higher end card.

Thanks for submitting more evidence to my point, you are clearly marking a trend where only the high end 1080 Tis can typically reach over 2000 MHz.

Oh and please submit videos from legitimate reviewers next time. Crappy quality videos of people running a single benchmark isn't a sign of a stable OC. I've had plenty of unstable OCs that worked perfectly fine in firestrike.

You talk about legitimate reviewers but you linked a Joker productions video, how ironic can that be lmao

I pointed you to videos where people actually manually overclocked their cards, also comparing gtx 1080ti overclocking potencial with a 130w card like the gtx 1660 is just ridiculous, because you know 1660 uses way less power and faces way lower limits. You can keep ignoring it and I will keep grabbing popcorn while reading every comment section from every article on this website with you defending AMD.

 

[Evernessince]

You talk about legitimate reviewers but you linked a Joker productions video, how ironic can that be lmao

I pointed you to videos where people actually manually overclocked their cards, also comparing gtx 1080ti overclocking potencial with a 130w card like the gtx 1660 is just ridiculous, because you know 1660 uses way less power and faces way lower limits. You can keep ignoring it and I will keep grabbing popcorn while reading every comment section from every article on this website with you defending AMD.

Yes and I pointed out the issues with the videos linked. The videos you link were supposed to be representative of your average 1080 Ti, which a lightning Z most certainly is not as it's MSI's most expensive high end card, and tested in a way that can be verified. That's why I cited silicon lottery as being a good source for 1700 IRL OCs. They sample 1,000s of chips and do so in a reproducible manner.

You should have at the least been linking the lower end 1080 Tis as your claim was that every 1080 Ti can reach. Instead you linked high end to ultra-entusiast models, which will clearly skew the results.

The biggest shame here is that there isn't any 1080 Ti overclocking database.

 

[144hzGamer]

Tbh every pascal gpu reach 2000mhz easily. Is beyond that that silicon starts to play a role. 2000mhz is pretty standard for pascal.

Btw techpowerup reviewed 5 different gtx1660s and all of them hit 2050mhz, with the msi gaming one hitting 2100mhz. I highly doubt any 1660 wont hit 2000mhz

 

[Evernessince]

Tbh every pascal gpu reach 2000mhz easily. Is beyond that that silicon starts to play a role. 2000mhz is pretty standard for pascal.

Btw techpowerup reviewed 5 different gtx1660s and all of them hit 2050mhz, with the msi gaming one hitting 2100mhz. I highly doubt any 1660 wont hit 2000mhz

I believe for Pascal temperature plays a big role as well. I remember watching a video on GamersNexus back near the 10xx series launch where steve made a statement to the tune of: You'll get max clock speeds at 69c and as you increase temp max clock goes down. That's a completely old memory though so don't quote me on that exact value.

 

[loki1944]

Considering I saw a 1060 3GB go for $76 and I got 2xXFX DD 390s for $92 apiece of ebay in the past month...definitely not impressive. If prices drop much more on used 1070s/980Tis, the 1650 will be an absolutely terrible option for bang/buck.

 

[Lightman]

The only gaming software I use is X-Plane 11...and in that application...RX cards are terrible (at least until the move to Vulkan).

Yes...the 570 is faster...but it's also a lot hungrier on power. I also like the idea of silent machines (Mac user for work)...and so the idea of fans noisily whirling away to disperse heat is a major turnoff.

The 1650 is not fast...but it fulfills a requirement I find appealing.

If my needs weren't so specific...I'd go for the 570...but as it stands...the 1650 comes across as quite attractive.

 

[Bluescreendeath]

So this article proves that the RX580 is and was the better option to the GTX 1060, yet Steam shows that the 1060 outsold nearly 15:1

Why are consumers such *****s.

You're forgetting that the GTX1060 was competing with the RX480 at launch, not the RX580. The RX580 was a RX400 refresh that didn't come out until a year after the 1060 released, and the RX580 had shortages too when it finally launched iirc. So the RX500 was 1 year later and had shortages.

If you take a look at the benchmarks between the 1060 and the RX480 when they first came out, they were pretty close but the 1060 still had the advantage in the forst few months after launch. The 1060 lauched at about 10 bucks more, but also used much less electricity and ran cooler.

If you remember, the RX480's low power models (150w) had a problem of drawing too much power from the motherboard PCIe and bricking lower end motherboards. They eventually fixed this by slightly lowering the clock and redistributing the power draw via software update, but people were buying the 2x power pin or 8 pin versions to be safe. Those more power hungry rx480s were drawing 180-220w max draw - which was singificantly more compared to the GTX1060 at 120w-130w max draw at roughly similar performance.

So if you look at the history, the GTX1060 was actually superior than what AMD had at the time - the RX480.