Anybody with 6 core CPUs? (Especially Intel)

First question and for curiosity only, what's your GPU?

I'm prety sure that your 2500K @ 4.5 is more than enough even to a 780 Ti in normal conditions.

Why are you running PCI-E @ 8x only?

You can't see the diference in performance from changing the AA level as you are seeing. 7fps in 100fps its not too much but 7fps in 35fps is a lot. Without AA you have 25% more performance and that's a lot so there's not a bootleneck there.

Before i Bought my 3770K i searched a lot in the internet and for my card i'm pretty sure that a I7 920@4.0ghz would be enough to every GPU available on the market so your 2500K @ 4.5ghz its more than enough too.

 

That RFactor 1 benchmark I believe was with the HD 6970 (currently running a 7970 @ 1170/1650). I had 3 6970s I trifired, but also tested in xfire and single GPU mode.

It was running PCI-E 2.0 because Sandy Bridge does not support PCI-E 3.0, and @ 8x because I had a PCI-E soundcard installed and that would drop my GPU to 8x, stupid Intel and only 16 PCI-E lanes.

I have upgraded to LGA 2011 X79, PCI-E lanes/bandwidth shouldn't be a problem anymore

What you say about the 25% reduction in performance may be true, you have a good point.

I need to do more testing!!!! Ahhhh!!! Lol

With my new build I will test GT Legends, RFactor 1 and Game Stock Car 2013 in single core, 4 core and 6 core mode, and RFactor 2 in 2 core, 4 core and 6 core mode, while making sure to use "AIMinPassesPerTick=10" in order to place much more load on the CPU.

Does RFactor 2 still have "AIMinPassesPerTick" for the A.I.?

 

Care should be taken when pushing older games to use more than a single CPU. There have been games where you actually have to force a single core only in order for the game to run correctly.

 

In case of NVIDIA, graphics driver may utilize available cores, regardless of how many cores game is using.

I'm not sure about AMD, but I'm guessing it's the same there.

 

Almost finished putting the new build together

 

I tested the Rfactor2 at 1280x1024 (53-70fps) and 1920x1080 (46-65 fps).The settings in game are very high . Have I a bottleneck with the GTX 760 ? (the CPU is a Q6600@3.2). Do you think that with a I5 4670K i would win more fps?

 

I would have expected a much bigger gain of fps from 1080p to 1280x1024, I believe you definitely have a CPU bottleneck there, unless you are using a huge amount of AA for those tests, are you?

 

if I remove the CPU OC (3.2GHz to 2.4GHz) only lose 5fps . What does that mean? If the problem is the should lose more fps...or not??

 

Addict, your CPU is the limiter on your system. Changing to an i5 will likely see you jump up to 65+ on the low side, over 100+ at 1280 X 1024.

 

Trouble is he's not obviously bottlenecked either way, he's actually getting FPS increases when he reduced resolution but on the other hand he lost FPS when he reduced CPU speed. He could slap a 780ti in and it would be better but it would be held back by his CPU. Conversely he could change to an i5 but if he's got the 760 well loaded up with very high settings then he's not going to see the full benefit of the i5.

Put the OC on the CPU back to 3.2GHz and reduce the resolution to say 1024x768, if the min and average FPS doesn't change or hardly changes from the 1024x1024 results I think I'd upgrade the CPU. If the FPS minimums are above 60 (the more above the better) I think I might upgrade the GPU.

Trying to find something that'll swing the decision more one way than the other, just hope you don't come back and tell us that at 1024x768 you're getting 59-60 minimums lol.

 

Ya try reducing the res to 1024x768 and then try re running the test to get a better idea of the CPU

 

Just to comparison i'll do some tests downclocking the 3770K to see where I start having bootleneck. After that i'll test my brother's 660 Ti Pe to compare with my previews result with Q6600 @3.6

brb

 

I will be attentive to the results

 

First results... Little diferent from what i expected... I did not downclock CPU because actually it does not work like my old X38 system... Instead i disable Hyper Threading and cores in bios.

I did the test with the same replay that its on other topic (R9 290 vs GTX770)



Before i tell my conclusion tell me your guys

 

My conclusion: You need to send your machine to me ASAP!

 

@Nuno

Just a little clarification please, you say you disabled cores and HT in the bios, yet your graphs are labelled core/threads, what exactly is 4 core/2 threads? When I ran an i3 (dual core with HT) I recall HT being an on/off setting I don't recall being able to turn it off per core, are you able to in your bios? I did wonder if you were using procmask or something but then you wouldn't need to go into the bios apart from changing the CPU multi (have you really no multis below 35)?

edit:

Just to make myself a little clearer (been told I don't always do this and assume too much lol), 4 core CPUs with HT like your 3770K are usually referred to as 4c/8t, 4 core CPUs without HT like my 3570K is usually referred to as 4c/4t and dual core with HT CPUs like my old i3-2120 is referred to as 2c/4t.

In your graph the bottom 3 labels make sense to me but the top 4 don't which makes me wonder if the bottom 3 are what I assumed they are.

 

Was a copy/paste mistake MJP.

Yes i can't run multi below 35 unfortunaly.

I updated the graph

 

I did wonder, now it's all clear, surprised you didn't run 4 cores with HT off though. I remember quite awhile back someone posted with quite a bit of info (a very decent spec pc too IIRC) and came to the conclusion that an i5 @3.6GHz was the most that was needed. I didn't have the hardware at the time to check for myself but doing a bit of testing and extrapolating I remember thinking he was pretty close.

The 2 core results are interesting re.the reduced minimums especially when you get some people (it usually seems to be AMD vid card users) asking why rF2 only uses 2 cores and I know I replied a couple of times showing 4 core usage (both when I had my i3 and now with my i5) then there was a post by a Dev saying that the Nvidia drivers could use more than 2 cores when necessary whereas the AMD drivers couldn't. I've no idea whether AMD vid cards quite literally can only use 2 cores or it's a maximum of a 2 core load spread across 4 (or more) cores as I've never seen anything relevant posted.

 

There's no special reason for not doing with 4 cores HT off. It spends a lot of time but tomorrow i can do that

Two cores its not enough... maybe with 4.5ghz OC in two cores I get same result that i get with 4 cores.

Tomorrow i check again

 

Thanks a lot Nuno!
While having 4 cores certainly helps improving that minimum FPS, it seems that with 2 cores you already get 78FPS which it is pretty good.
My question is this now: with 2 cores without hyperthreading can you race smoothly online with 25 cars, without too much stutter?
I ask this because I am guessing (without knowing and without testing) that replays use less CPU than online multplayer races and I am seriously considering buying an i3-4330 3.5Ghz 2c-4t computer because I am very low on money, but I am thinking that maybe waiting a bit more for an i5 could be worthwhile... but the less I must spend the better.