• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

NVIDIA 4000 Series

@Curtis vincent is a scalper, pure and simple. He's trying to make *money* on it and not do anyone a favour, my definition of one anyhow. I mean don't get me wrong, nothing wrong with it imo did it a few times last gen (had a play then moved it on usually for a bit more), paid for a few beers :)
 
selling one card hardly makes him a scalper, seems like more of a "buyers ramose" and becoming an opportunist

according to Einstein @puragtory there is no shortage anyway, so I don't get why people care
 
Last edited:
I spoke to @Curtis vincent yesterday (before this all became a thing) regarding his sale and he seemed just to be someone who had changed his mind post AMD launch, I know it’s not great to try and add a bit more on top of rrp but I completely understand the temptation (and if you don’t then I don’t believe you). The reaction on here to it has been a bit hyperbolic imo.
 
MidKnight said:
I spoke to @Curtis vincent yesterday (before this all became a thing) regarding his sale and he seemed just to be someone who had changed his mind post AMD launch, I know it’s not great to try and add a bit more on top of rrp but I completely understand the temptation (and if you don’t then I don’t believe you). The reaction on here to it has been a bit hyperbolic imo.
Click to expand...
But he hadn't spoken to anyone via DM to sell?

Either way, it's been dealt with now.
 
Chuk_Chuk said:
I’ve just rewatched this and considering all the big talk about double the number of samples per scene, the first two benchmarks are pretty disappointing imo. The Maya one was closer to what I was expecting.
Click to expand...

He does explain the results. Double performance on tests with lots of reflective surfaces. Not double in the mat scenes and the animation where there is a lot of CPU initial set up time (ie half) included.
 
Bill Turnip said:
But he hadn't spoken to anyone via DM to sell?

Either way, it's been dealt with now.
Click to expand...
No - but I DM'd him to ask him if he could point me to his eBay listing (and asked him why he was selling etc - and his answer was very plausible) - I took a look and decided no, as I mentioned I am without a GPU and had a bit of a moment of weakness yesterday and had a look around to see if there were any fairly 'reasonably' priced ones but then I gave myself a slap and have just decided to wait for either an FE or an Asus.
 
MidKnight said:
No - but I DM'd him to ask him if he could point me to his eBay listing (and asked him why he was selling etc - and his answer was very plausible) - I took a look and decided no, as I mentioned I am without a GPU and had a bit of a moment of weakness yesterday and had a look around to see if there were any fairly 'reasonably' priced ones but then I gave myself a slap and have just decided to wait for either an FE or an Asus.
Click to expand...
I will gladly help out a genuine user (who receives, opens and uses it). Not someone that will need to sit on it and thinking about it.

It’s my skip generation anyway, so can help with any warranty matters.

I had the FE in the basket a number of times but decided it’s not for me.
 
Trickle said:
He does explain the results. Double performance on tests with lots of reflective surfaces. Not double in the mat scenes and the animation where there is a lot of CPU initial set up time (ie half) included.
Click to expand...
There are a number of issues with both his explaination and yours.
Scenes with reflective surfaces need more bounces and samples to achieve a clean render in comparion to the same scene without those reflective surfaces, hence they take longer. The difference between gpus is simply how quickly they can calculate the bounces. Unless the 3rd gen RT cores have systems in place to specifically speed up recursive bounces in reflective surfaces only (or deep recursive bounces in general), we should be seeing a consistent speed up in render times regardless of material used.

About your explaination on CPU initial set up time. Set up time is dictated by the following
The number and total size of textures in the scene
Polygon density of models
Complexity of shaders (how many nodes does it use)

Render size does play a small part but it is tiny compared to the above I have listed.

The junkshop scene isn't a particularly demanding scene for the CPU to set up. It took 20 seconds on my system using a 5900x and a sata SSD, at 8k which is the same amount of time as a 1080p render. Which is a tiny amount of time compared to the total render time he quotes of 30 ish minutes. So that doesn't explain the discrepancy.

The junkshop scene is present in both his benchmark and in the blender benchmark. The blender benchmark results that he shows has the 4090 at double the sample per minute of the 3090, so in theory you should get half (or close to half) the render time.

Edit: Look again at the first frame results for the animated frame. The difference between the 4090 and 3090 is almost negligible. That does appear to take 1-3 minutes to setup, however I would not expect the results to be so close between the 4090 and 3090. Frame 48 render time is still not particularly impressive.
 
Last edited:
brostradamus said:
videocardz.com

NVIDIA GeForce RTX 4080 gets tested in Geekbench, 30% to 37% faster than RTX 3080 - VideoCardz.com

NVIDIA RTX 4080 tested in Geekbench Finally, some independent tests of RTX 4080 GPU. With just 9 days until official launch of RTX 4080 graphics cards, reviewers should now have their cards in hand. This obviously leads to first performance leaks surfacing before the embargo. According to newly...
videocardz.com videocardz.com
Click to expand...


It's only testing Cuda performance. It shows the 4080 is 15% ahead of the 3090ti but still 50% behind the 4090.

What's more relevant though is 3D benchmarks and the 3D Mark Timespy results have been released as well. In Timespy extreme the 4080 is 35% ahead of the 3090ti and 40% behind the 4090.

These benchmarks make the 4080's price look bad as the performance per dollar is higher on the RTX4090. For example the 4080 generates 11.8 points in Timespy Extreme for each Dollar spent where as the 4090 generates 12.5 points for each dollar. The 4080 needs a $200 price cut.

videocardz.com

NVIDIA GeForce RTX 4080 3DMark TimeSpy scores have been leaked as well - VideoCardz.com

NVIDIA RTX 4080 tested in Geekbench Just hours after the first RTX 4080 scores made their way to Geekbench site, the same unreleased SKU has now been tested with 3DMark software. Hardware leaker MEGAsizeGPU, who has an excellent track record recently, is now the first to share RTX 4080 3DMark...
videocardz.com videocardz.com
 
Last edited:
ExcitableBoyUK said:
can’t see the next batch of FE’s getting scalped which is good news. There are so many just sat unsold on eBay.

Do you even get a warranty if you buy from eBay?
Click to expand...
Depends on the vendor. Some do not recognise 2nd hand warranty at all (example is NVIDIA with FE cards - they specifically state that on their website), some other do not care or don't ask. Best to check particular vendor's terms of service about it or forums.
 
You must log in or register to reply here.
Back
Top Bottom