List of Features/Improvements

[Art48]

Hi,

I will post my personal list of improvements/features/ideas here regarding the Benchmark.
Let me preface this that I was suprised I didn't have to install anything. Positively surprised. I liked the time you guys aimed for (4-5min as I read) and it is definitely good as in not taking too much time.

1.
When I submitted my statistics I can't find any way to go at that position where I am (let's say place 300) and find my own statistic there. Would be cool if there's some kind of jump to button. Maybe show me my user name in full on the list when I'm logged in.

2.
I used the benchmark mainly because we are looking at buying another workstation soon and ideally there was a filter menu where I could say "Which Hardware would I need to cut render times by roughly 50%?"
Having some kind of comparison side by side function would be cool. You can improve that even further by tracking prices maybe on amazon or something and provide me with a function like "How much would I have to invest for a 25% performance gain?" I would also be willing to use a chaosgroup affiliate link or something for buying a new processor based on the benchmark results since you don't seem to make money from the benchmark and this might nurture future features etc.

[daniel.taskoff]

Hello Art48,

Thank you very much for taking the time to share your opinion on V-Ray Benchmark with us.

1. After submission, you should've been opened an url of the form https://benchmark.chaosgroup.com/cpu?id=3 with id=3 being your hardware configuration's id. Opening such an url will show the page containing the highlighted result.

2. We'll definitely expand the filter/search functionality, but I'll refrain from giving details, leaving that to someone who can give more concrete ones.

Best regards,
Daniel

[maxwood]

It would be nice if we could filter for single CPU/GPU results only to make comparisons a bit easier. I think the biggest request from me personally would be to improve the search function so that it shows all of the results for your search criteria instead of just one.

[Jiri.Matys]

I agree. Filtr all lines with hardware I choose is important.

[vlado]

Now that there is actual data, we (or anyone really) can do all kinds of analyses of hardware for price vs performance. We definitely have some ideas in that regard.

Best regards,
Vlado

[biochemical_animations]

Now that there is actual data, we (or anyone really) can do all kinds of analyses of hardware for price vs performance. We definitely have some ideas in that regard.

Best regards,
Vlado

Yep, me too...good stuff.

[im.thatoneguy]

It would be great if you detected actual frequency instead of just the product name. I ran the test and scored comparably to one 7700K. Booted up with a 15% overclock and matched another score exactly.

It appears to group by CPU + GPU (which seems unnecessary since the two are independent). So when I posted a slower score it told me I was part of a group that was faster than mine (presumably due to water cooled overclocking).

It would be great to not just see the fastest score but a stock score and a distribution of other scores. Maybe with notes too.

2:36 - @4.2GHZ Stock
2:12 - @4.64GHZ "15% overclock, air cooled
2:13 - @4.65GHZ "overclocked with tower cooler."
2:01 - @5.00GHZ "water cooled"

As it is, nearly every score at this point appears to be an overclocked score at some unknown level not the stock cpu.

[vlado]

We plan to put more information about the actual clock speed.

Best regards,
Vlado

[Sprawl]

Hi Vlado,
great idea to have a common benchmark.
It would be great if the scene for GPU and CPU should be the same, so it'll be possible to misure the difference between CPU and GPU.

Thank you

Alex

[vlado]

It would be great if the scene for GPU and CPU should be the same, so it'll be possible to misure the difference between CPU and GPU.

We will not be doing this for the time being; the difference is too large for different scenes and use cases and we do not want to mislead our users.

Best regards,
Vlado

[rajdarge]

sorry I started a new thead - actually wanted to tack it on here

So I notice this nice little thingy in the newsletter.
It runs really slow on my machine - to be expected, old q6600 with music playing, downloading crap and open web browser.
But after 3 minutes I stops.
I was just trying it out, but I expect it will take 30 min to finish. I don't have 30 minutes.
Can it be a little faster? (eg smaller image).

However when I click on "statistics" it takes me to a leader board page.
Can I believe the results?
all the top scores are for the Xeon CPU's - the more cores the better, regardless of clock speed almost.
most of the i7/5/3 don't even rate, or show up?

I am due for an upgrade on my single licence machine, and i was thinking of a 10 core i7 i-6950x
should I just opt for a slower base speed and something like E7-8890 V4 (if I can afford it)?
So are the Xeon's the way to go for a single machine rendering engine?



Raj

[varosi]

Current version of V-Ray Benchmark is doing rendering of fixed scenes, so you could compare the results between different hardware. We'll think for next versions about how we could benchmark fast even greater range of processors.

You may thrust results. They are counted on the rendering part of ray-tracing the scene. They could differ depending on side effects like processor frequency throttling (usually on laptops), programs and services that run in parallel with the Benchmark, overclocking of the processor and memory. Xeons have usually a lot of cache memory which is a great factor, too.

We cannot tell you what processor to buy, but you could use V-Ray Benchmark results to help make the decision for you about rendering with V-Ray. Keep in mind that this Benchmark give results for concrete scenes and there could be variations in performance if you render different scene (with different features).

[rajdarge]

As the cinebench and the 3ds max benchmarks are the ones i have used before - its great to see something vray specific. But i look at tables of other peoples machines to compare them to what my macine can do on sites such as toms hardware, anandtech etc. So i do yhis to work out if its worth upgrading at all, an if so to which machine. I usually never trust a single site or benchmark so its also good to see the gpu and CPU measured.
Sure its a marketing tool to promote vray, but it has real practical uses too

[varosi]

Every benchmark tool has some specifics and you're doing absolutely right to test with different tools. The best machine for the money should be machine that is doing cost efficiently for tasks that you'll need it for.

[JD3D]

It would be nice if we could filter for single CPU/GPU results only to make comparisons a bit easier. I think the biggest request from me personally would be to improve the search function so that it shows all of the results for your search criteria instead of just one.

I second this - I'm about to buy another couple of GPUs, and it would be great to be able to select each GPU and see how it compares against another!