Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
The Athlon 3000G is due to launch on Nov. 19. If you're looking to build a super budget PC that should do the trick. The CPU MSRP is $50US. You can get a B450 MoBo, there is absolutely no need to get X570 as PCIE gen 4 simply isn't needed, for less than $100. 16Gb of 3200 RAM is 60$ and you can get 64Gb for right at $120, more RAM is better in a build using a iGPU, a 2Tb HDD is $50, an 80+bronze 500W PSU, gives you plenty of capacity to add a discrete GPU, is $45, You can get a case for $40. If you already have a Windows license that means an entire PC, won't be great at rendering but it will work until you get a graphics card for $400. If you shop the weekend after Thanksgiving you could even shave a few bucks off that as memory, at least will likely be on sale.
It doesn't have a very huge difference. the performance difference between 1060 6GB, and 1660 is roughly 13%. Ti is again 13% faster than 1660, and 1660 super is in between of 1660, and its super version. So TI is not a good investment. you can easily achieve that performance with 1660 by overclocking.
No, you can't. First you cannot pick up 6 or 7% performance by a GPU OC except under subambient cooling. Second iRay really only cares about CUDA count and clock speed is of very little importance.
Would any of you have some insight on a EVGA Nvidia RTX 2070 8 gb? I am considering upgrading to this from an EVGA NVIDIA GTX 970 4 gb ...
Clock speed does matter. Downclock your cards to 50% and see how that goes, I guarantee it wont be pleasant. Rendering is a combination of CUDA and clockspeed as clockspeed directly effects how much work is done.
The real question is how much performance can overclocking have, and that depends on every card as it can vary. I do agree that overclocking for Iray is probably not a good. Iray is stressful enough as it is, overclocking would only add to that. But to be fair, gaming can easily be even more stressful than Iray. I know. My temps with Iray on my main GPU will be 70C, but with gaming it can hit 80-84C. That's because I like gaming at high refresh rates, so even less demanding games can run the GPU harder because of that.
Anyway, overclocking might knock a couple minutes off an hour render. Probably not all that worth it. And I don't think a 1660 can just be overclocked to match a 1660 Super or ti. And even if you can overclock it to match them, consider that the Super and ti could also be overclocked, which would regain that advantage all over again. The Super and ti are very similar, so if looking for one, just get whichever is cheaper.
The 2060 Super is right at the 2070 in performance, so get a 2060 S instead. Unless you are talking about the 2070 Super, or a 2070 is a at really good price less than a 2060 S.
Eh, these naming schemes are ridiculous. Supers and tis everywhere, it is all needlessly confusing.
Here is what you can do. There is a benchmark in my sig, download it, and compare your 970 to the times on the chart. The 2060, 2060 Super and 2070 are all on there, plus more. There is no 970 on there yet, so if you do take it, please post your results like other users have. I bet the difference is pretty big, I used to have a 970.
One thing to note is that exact performance of RTX cards can vary depending on the scene makeup. CUDA cores are very predictable with how they scale, while RT cores work better with more complex scenes. So depending on the types of scenes you create, RTX may help more. Think of RT cores as sort of a bonus on top of the CUDA cores. But make no mistake, they can really speed up a render over a card without them.
BTW, the next upgrade of GPUs is set for mid 2020, with rumors suggesting June/July. Though nothing is officially confirmed by Nvidia, I am just letting you know. I personally think waiting may be worth it, as I expect big things from the second generation of RTX cards.
I am buying the AMD Ryzen 7 2700 in December & the Radeon RX 570 8GB in January. That will make for an desktop excellent PC at about $620. Later adding 3 sticks of 1x16GB DDR4 RAM & an 2TB NVMe SSD will jack the price of the desktop's components of to about $970 but most of the components are components you don't replace frequently like is done with GPUs & CPUs..
I've already bought the desktop case & incidentals in November. Then after I come back from overseas next August I will be adding a Zoltec RTX 2070 Mini for $400 or so and if available and somewhat cheap, AMD Ryzen 9 3950X. Those two will again send the compute power through the roof of what I am building now but not near as much as from what I'm upgrading from. However, those last too are dependent on the new nVidia Ampere models and Generation 4 AMD Ryzen CPUs not changing my plans but those look to be the best compromise between current modern compute power & price.
So you can spend $250 in one month but can't save one month to save $250 and have $500 on hand to buy something like a Zoltec RTX 2070 mini going for $399 as I type? That's less than $450 including taxes and shipping. That's not six months, not by a long shot. It's not like the number system goes 1, 2, 3, ..., 247, 248, 249, 250, 1,000,000
Just curious ...is Zoltec a reliable brand? I can get a Zoltec nvidia 2070 much cheaper than an evga one but I was concerned on the performance of them
Thank you for the information, I will try and run that benchmark test
Never heard of Zoltec. Zotac is a great brand. Zotac is particularly well known for making the smallest cards on the market.
sorry spelling error I meant Zotac ... thanks
I am hoping for some good Black Friday deals on GPU's but regardless I will upgrade now, I can't seem to use half the products I have purchased such as anything dforce ( especially hair) or HD as it drops to CPU and takes forever to render so I have mostly been doing just some basic portrait renders and learning how to use some of the utility products I have purchased. Thanks for the info on the different options, I appreciate it!
On the contatry, you can. Turing GPUs are very efficient; hence you can overclock them to gain a higher performance. It's proven over and over in many youtube videos. It might stress the card, but you get 3 years warranty from major manufacturers. Zotec gives 5 years warranty. Second, the CUDA count is not the sole factor for improving the rendering time. You need a higher base speed, memory clock, memory speed. 1660S has a higher factory base clock than its TI version, its memory speed is also higher (14Gps), boost clock speed is also high in both 1660 versions (not TI). So nowadays getting 1660S is much better compared to 1660, and 1660 Ti. Ti is a waste of money. You better invest for 2060 if you can spend extra for 1660 Ti version. For someone who said Ti can also be overclocked. It was done by a youtuber, but its performance gain is little compared to an overclocked 1660 version. It's not really worth. You spend like $30+ for getting a little bit of performance when getting a 1660ti from 1660S version. People who look for a budget card want a cheapest capable card that can render their scene. If they have to spend more, it defeats the purpose of getting a budget card. Why not go straightly for Titan RTX if you want the best card in the market. 1650S is also good, especially with scene optimizer, but since it has just 4GB VRAM, it's not a best budget card.
If someone is looking for a new, budget GPU for iray rendering, My recommendation is getting a 1660S version. If 1660S is not available then 1660. Both are good.
Zotec gives 5 years warranty. So it's a reliable brand.
Watching YT will definitely not teach you how to OC or what gains can be had.
For instance 1660 Super and 1660ti are not just different branding of the same underlying GPU chip. The 1660 Super is TU116-300 while 1660ti is TU-116-400.
The reason there are bunch of differences in memory performance between the 1660 and 1660 Super is not because the memory has been OC'd in the Super. The 1660 has GDDR5 while the GDDR6.
Now on getting 5+% performance gain out of any Turing architecture GPU by overclocking. Turing like Pascal before it will down clock to maintain ideal operating temp, you can prove this for yourself go in and set the card fans speeds to 1000rpm and run a stress test then set the fans to 1100rpm and run the test again. The 1000rpm and the 1100rpm runs will have essentially the same temp but different frequencies. So running an overclock, which means increasing the voltage and therefore putting out more heat, will net you next to no performance gain without using at least water and to get the sorts of gains you're talking about you'd need the OC'd card to run much cooler than the non OC'd one which would require at least chilled water. Overclocking GPU's used to be a thing, Before Maxwell, but now? You likely won't even be able to get the results the factory OC cards have, since those are binned chips. The reference 1660 Super reports a boost clock of 1785 Mhz while the factory OC's I could find max out at 1830Mhz which is a 2.5% OC. IMO that's not worth spending the money on the factory OC and certainly not worth doing at home unless overclocking is your hobby.
Yeah right I should take some random stranger's comment over a professional's advice.
Still there is a difference in specs. Ti version has more cuda cores, base clock is low. memory speed is high in S version, badnwidth is also high.
Nobody said S version is overclocked; hence S has more performance. The memory speed is throttled due to lower GDDR5 memory in 1660. According to many reviews, 1660 is 13% slower than Ti version, and 1660S is in between the base version, and the Ti version, meaning the difference between S and Ti is now very insignificant.
Completely wrong. "Time Spy garners a 15% increase in performance while Firestrike sees 12%. The interesting part is that power draw only increased by 8% and thermals by 5%, a very worthy trade off for the performance that you net." - wccftech
Saying 5+, even with the plus sign undermines the value of 1660 and its S version. So guess what. 15% increase in performance mean, you essentially reach the base level of 1660 Ti version suppressing the super version too. So there is no point of spending $30 for TI version, but it's worth to spend $10 for Super version, as it has GDDR6 memory which prevents throttling caused by lower memory bandwidth, and it's also almost fast as Ti version. So you get roughly the same performance card for extra $10 without needing to overclock, but if the money is a problem still going for 1660 is perfectly okay, because the difference between base, and S version is not much. Ti version might have 128 more CUDA cores, but it's not worth to spend extra $20~30 for getting the extra CUDA cores when you can have roughly the same performance with S version. Besides, the amount of CUDA cores alone doesn't improve the rendering time. Sure, you can overclock TI version too, but overclocking Ti doesn't gain a good performance as non T version, it's probably because of what you said. also when you render content, 13% won't do a huge difference, especially when reducing the resolution of textures, which most budget content creators often do. If you intention is to improve the render time significantly, then better invest for RTX series instead of GTX 16 series. The only reason someone checks 16 series is because they don't want to spend a lot of cash for a graphics card, but here in this thread the first recommendation is to get 1660 ti version, which completely defeats the purpose. seriously, someone asked is 1660 good, but you said it's better than 1050 ti, and get 1660 ti and RTX version? How come 1050ti was compared to 1660? of course 1060, 1660, 1650 all are better than 1050Ti.! why recommended to get Ti, or RTX version when OP signals about the money problem by saying "they're way cheaper than most Nvidia cards". This is what I don't like about this forum. Instead of explaning the advantages, disadvantages, use cases of the card they intend to buy, they are pushed to buy more expensive ones because people think they can afford it. If people have enough money, they don't have to waste time on checking benchmark results of graphics cards, ask questions in forums, they can straightly go for Titan RTX. When you are on budget, even $1 matters. So if you recommend someone to spend more, you are expected by people to explain why they should spend more instead of just dropping more expensive card list in the thread.
I don't want to argue and waste time here. My recommendation for people is not take my or anyone's word, just check some of the youtube videos and see the performance difference between 1660, S and Ti versions, then you would know getting either one is fine. If you need more performance, then go for RTX series, but it will cost more.
!
There is no base version when talking about the 1660, 1660 Super and the 1660ti! I know Nividia has not made this clear but these are not 3 versions of the same GPU chip. The 1660ti is an entirely different chip. The difference between the 1660 and 1660Super is mostly due to the change from GDDR5 to GDDR6. GDDR5 and GDDR6 are different VRAM modules. GDDR5 is not throtteled or any such thing. GDDR5 runs at very well known speeds and the GDDR5 in the 1660 is running at those speeds.
As to watching vids touting huge performance boosts from OC's, especially stuff by guys like wccftech, just don't. There are legit and well respected tech channels on YT. If you care about OC's and water cooling JayzTwoCents is a good channel that does actual overclocked builds and discusses performance before and after and you can see from his videos that after GPU Boost 3.0 became the standard on Nvidia GPU's overclocking the cards became mostly pointless. Or you can watch GamersNexus who approaches computer parts from a technical perspective and does some overclocking (not on GPU's because no one bothers anymore). Or for people who are serious about professional and enterprise computer hardware there are folks like Level1Techs and ServeTheHome.
As to your claims, you're welcome to prove to yourself how little GPU clock matters. Run an iRay benchmark, down clock your GPU and then run it again. Compare your times. A 5% DC will not get you a 5% increase in render time.
https://www.youtube.com/watch?v=-hWMPBrrKn4&t=4s
My own opinion; I own a 980ti, which has 6GB.
It isn't enough.
Although - now I render in Blender's Cycles, it's fine as my Threadripper is faster and of course I don't have RAM issues with the CPU.