#4483836 Posted on: 12/20/2012 09:03 PM

---

Here's my ubber system

GPU - AMD Radeon HD 6320 (E-450 APU)
CPU - AMD E-450 APU
RAM - 8GB DDR3 @ 1333MHz

531 - Kitty Preset (576p)


433 - Custom (Kitty Preset at 720p)


410 - Cat Preset (720p)


243 - Tiger Preset (1080p)



I'm sure you all envy my gaming machine

In a more serious note...

I could not run any preset over cat with the cards and os i am on atm.
I needed to set the cat profile in ni and play to sli work more ...before score was only 755

Your CPU score is messed up also. That i5 of yours is only getting 50 on CPU score and mine is getting 35. Others are getting several hundreds on their CPU score. Also... Windows XP? Seriously? That OS is not taking full advantage of the hardware, for sure.

#4483847 Posted on: 12/20/2012 09:26 PM

---

those who scores 1000 on cpu bench what's your framerates at cpu bench?
mine starts with 12, ends 7 fps, does the benchmark use avx for physics or what?.. i hardly believe an ivy 6x better than a c2q..

#4483872 Posted on: 12/20/2012 09:55 PM

---

those who scores 1000 on cpu bench what's your framerates at cpu bench?
mine starts with 12, ends 7 fps, does the benchmark use avx for physics or what?.. i hardly believe an ivy 6x better than a c2q..

I can post mine but in the morning, rooms too warm now and i don't want to punish gpu any more, between this and FC3 the poor fella doesn't know what day it is.

#4483878 Posted on: 12/20/2012 10:00 PM

---

Win 7 x64 SP1 with i5-2500K@Turbo Boost and GTX 480@reference-stock 310.70WHQL (HQ)

Tiger preset:


Kitty preset:


Please click these screenshots to enlarge.

#4483887 Posted on: 12/20/2012 10:10 PM

---

Yeah, looking at your posted screenshot it seems you are scoring just over 1k with a 4.7Ghz OC and no HT, whereas I'm scoring just over 900 with a 5GHz OC and HT? clearly these results require further interpretation. I'm guessing it's either due to that test being heavily memory bandwidth sensitive, using some specific IB only feature or the posts about physx being implemented in part may further explain this.

#4483893 Posted on: 12/20/2012 10:15 PM

---

leave the hyperthreading outta the equasion (i doubt it's using it), only really two things, faster ram & the ivy vs sandy performance gap, it must be those two things, or one of them, perhaps the ssd makes a difference, this cpu is also run 24/7 @4.7ghz without any powersaving. gets a 7.9 in cinebench currently.
Apart from that i dunno.
what do you get in cinebench with hyperthreading disabled?

#4483915 Posted on: 12/20/2012 10:28 PM

---

those who scores 1000 on cpu bench what's your framerates at cpu bench?
mine starts with 12, ends 7 fps, does the benchmark use avx for physics or what?.. i hardly believe an ivy 6x better than a c2q..

Yeah..

I think its some lame Kepler+physx trick (probably something like with 3dmark vantage). Cowie with i7 had only 30? points because of 8800gts.

#4483920 Posted on: 12/20/2012 10:33 PM

---

Gen2 vs Gen3 also. This rig is full blown gen3 (except the ram is a bit slow, 2666 /2800 is more suited).
just trying to help, hope you know that, i know there's many ultra uber rigs here too.

#4483921 Posted on: 12/20/2012 10:37 PM

---

leave the hyperthreading outta the equasion (i doubt it's using it), only really two things, faster ram & the ivy vs sandy performance gap, it must be those two things, or one of them, perhaps the ssd makes a difference, this cpu is also run 24/7 @4.7ghz without any powersaving. gets a 7.9 in cinebench currently.
Apart from that i dunno.
what do you get in cinebench with hyperthreading disabled?

Your score for the last part 'loading times' is also 50% higher, even though I had it on an ssd, also I'm note sure but that last test might also be some sort of cpu based test, which seems likely given that I have it on an SSD. I'm going to wait and see if any of this is related to physx, especially as my 3dmark11 default driver scores are comparatively high.

Also I haven't tested cinebench with HT disabled, though is does lead to a large drop in score, and it's also very memory sensitive. I'll post a score in a short while, though my per core score for real temp's xs bench is around 2050.


edit:

I don't think pci-e gen2 vs gen3 should make much difference here as the fps isn't that high and there no multi-gpu data transfer bottleneck. I could be wrong.

Here's my cinebench score at 5GHz no HT, also the memory is only running 1600MHz but I'm still scoring 5.0.

#4483930 Posted on: 12/20/2012 10:47 PM

---

Gen2 vs Gen3 also. This rig is full blown gen3 (except the ram is a bit slow, 2666 /2800 is more suited).
just trying to help, hope you know that, i know there's many ultra uber rigs here too.

If it were a true cpu test it wouldn't matter, gen3 or gen4 for that matter

I think its just some flawed physx trick, again. In the end its not a real cpu test since its also affected by resolution and type of gpu you use..

But then again its a beta benchmark, so i won't judge it.

#4483934 Posted on: 12/20/2012 10:51 PM

---

Basically the z77 platform i'm referring to when saying gen3, and all the features this includes.
But sure it's still beta.

#4483937 Posted on: 12/20/2012 10:53 PM

---

Here's my cinebench score at 5GHz no HT



you score 8.00 with your 5ghz, i score 7.93 with my 4.7ghz
sry double post.

#4483948 Posted on: 12/20/2012 11:04 PM

---

you score 8.00 with your 5ghz, i score 7.93 with my 4.7ghz
sry double post.

Yes, with 2.4GHz vs 1.6GHz memory though, still your catzilla cpu score should not be that much higher than mine, it should be similar, assuming HT is not used.

Also, here's my score with HT.



So yeah, it seems there's more to it.


Edit:

correction on the loading times test, it's measured in ms and that means my score is actually 50% higher. Which further complicates matters.

#4483983 Posted on: 12/20/2012 11:35 PM

---

Yeah i seen that your loading time was excellent.

#4483990 Posted on: 12/20/2012 11:42 PM

---

Just noticed that someone with a 2500K and a single 680 scored almost 20000 points. Looks like i have some work to do lol

#4483992 Posted on: 12/20/2012 11:43 PM

---

ok, new run...looks better now, and didnt even tweaked m ysystem or closed firefox, etc..

dunno why my first run was soo low on CAT!! (3245 or so)
Now on TIGER, yes, tiger with downsampling, i have 4066..

Big Diff, i have no clue what ive done, here..
anyway, heres my result for Tiger, the rest coming soon..
--------------------------------------------------------------------------------

TIGER: (Downsampling)

Klick Mé... POINTS --> 4066

Also added temps and info screen, Nice cooling hahah

Cat:

Klick Mé... POINTS --> 6730


Kitty:

Klick Mé... POINTS --> 8336



-BetA-

Edit.... Ha, @ -Tj- did beat yaa in CPU Test..
Fur Test is also close, , to close when i consider u have a way more powerfull Card..

#4483999 Posted on: 12/20/2012 11:46 PM

---

ok, new run...looks better now, and didnt even tweaked m ysystem or closed firefox, etc..

dunno why my first run was soo low on CAT!! (3245 or so)
Now on TIGER, yes, tiger with downsampling, i have 4066..

Big Diff, i have no clue what ive done, here..
anyway, heres my result for Tiger, the rest coming soon..
That CPU score seems pretty poor for that system... although who knows. This benchmark scores strange results anyway.

#4484037 Posted on: 12/21/2012 12:43 AM

---

Any ideas why my Fur score is so low?
(it's not heat, max was 62c)

#4484047 Posted on: 12/21/2012 12:59 AM

---

you score 8.00 with your 5ghz, i score 7.93 with my 4.7ghz
sry double post.

You scores seem really low for the frequency you are running.

4.8GHz 3930k I get 13.91

#4484069 Posted on: 12/21/2012 01:45 AM

---

Those were 4c/4t scores and you're running 6c/12t. With 4c/8t I score closer to 10, with faster memory it could get closer to 11.

#4484071 Posted on: 12/21/2012 01:48 AM

---

TIGER: (Downsampling)

Klick Mé... POINTS --> 4066


-BetA-

Edit.... Ha, @ -Tj- did beat yaa in CPU Test..
Fur Test is also close, , to close when i consider u have a way more powerfull Card..

And you have 300mhz higher cpu clock ;D


About fur idk not really, but ok if it makes you feel any better

Edit:

run gpu on stock and then compare, your Oc'ed to the max gpu might come close but that about it hehe

here is mine on gpu stock clock

1st try


2nd try, this time with prefer max. perf and threaded opt on instead of auto


as you can see quite inconsistent. On 2nd try i got higher overall score, but fur mark, gpu mark and cpu mark is lower? doesnt quite add up to the final score.. O yea and I had basic theme by both.

#4484079 Posted on: 12/21/2012 02:03 AM

---

Hello Everyone!

You need to take my word. I'm from the dev team that made catzilla. If you have some questions about it's internals then go ahead and ask. We are currently doing some heavy testing and we are still fixing a lot of bugs that started to pop.

First of all. In a couple of hours you won't be able to submit scores on that version of BETA hosted here. Everyone who registered on our site will be informed about new download link for new BETA.

Second, we are sorry, that someone from ORB team has opened the payments tab for a moment, there were 10 people that bought catzilla already . Sorry for that inconveniance - you probably already got informed via email, that we will be sending your money back.

I would like to encourage you boys & girls for submitting the results to our page. Because of that we can see what reading are incorrect and we will fix them. Currently the CPU clocks are just HFM, but from new version we will take measurements during heavy loads, so this should result in proper overclocked values.

Also we store info about SLI & crossfire on the server already but we don't display them on he results page. You can view them by checking system configuration in each users tab.

We are not quite sure about NVIDIA clocks thou. AMD GPUs should have correct clocks. If you could please send resulltts especially for low end systems like old XP machines, Vista netbooks and especially 32 bit systems then it would really help us.

Don't worry we just send HW info to server and results. We install a kernel driver during installation process to have proper HW info. it's just done once and requires admin rights. 3d mark does similar thing.

As for common questions regarding PhysX. We just use CPU version. Spending GPU time on physX is a waste when you have fast CPUs.

Thanks a lot for your tests!

Michal 'bonzaj^plastic' Staniszewski

#4484084 Posted on: 12/21/2012 02:14 AM

---

Good to hear from one of the devs and i'm sure there will be alot of questions

On a modern vga and os now First try on my 680/2600k rig

#4484087 Posted on: 12/21/2012 02:20 AM

---

Hello Everyone!

You need to take my word. I'm from the dev team that made catzilla. If you have some questions about it's internals then go ahead and ask. We are currently doing some heavy testing and we are still fixing a lot of bugs that started to pop.

First of all. In a couple of hours you won't be able to submit scores on that version of BETA hosted here. Everyone who registered on our site will be informed about new download link for new BETA.

Second, we are sorry, that someone from ORB team has opened the payments tab for a moment, there were 10 people that bought catzilla already . Sorry for that inconveniance - you probably already got informed via email, that we will be sending your money back.

I would like to encourage you boys & girls for submitting the results to our page. Because of that we can see what reading are incorrect and we will fix them. Currently the CPU clocks are just HFM, but from new version we will take measurements during heavy loads, so this should result in proper overclocked values.

Also we store info about SLI & crossfire on the server already but we don't display them on he results page. You can view them by checking system configuration in each users tab.

We are not quite sure about NVIDIA clocks thou. AMD GPUs should have correct clocks. If you could please send resulltts especially for low end systems like old XP machines, Vista netbooks and especially 32 bit systems then it would really help us.

Don't worry we just send HW info to server and results. We install a kernel driver during installation process to have proper HW info. it's just done once and requires admin rights. 3d mark does similar thing.

As for common questions regarding PhysX. We just use CPU version. Spending GPU time on physX is a waste when you have fast CPUs.

Thanks a lot for your tests!

Michal 'bonzaj^plastic' Staniszewski

Thanks for posting, bonzaj.

I have a question. In this current version it seems Nvidia graphics card based systems are scoring much higher than AMD counterparts, where as in other benchmarks like 3dmark 11 the same is not true. Is this going to be fixed and will the benchmark properly balanced in the next version?

Also, thank you for clarifying that GPU Physx is not in use.

#4484090 Posted on: 12/21/2012 02:23 AM

---

Good to hear from one of the devs and i'm sure there will be alot of questions

On a modern vga and os now First try on my 680/2600k rig


I still find that cpu score very suspicious, why would 8800gts hider cpu score that much, doesnt make any sense at all..

With IvyBirdge & 8800gts you had 50 points but now with SandyBridge & 680gtx a whopping 1100+ points? heh xD


Do you have a 500 series gpu too? If yes could you please test it with your SandyBridge cpu?


Edit: And could you test 310.70 driver with current SandyBridge + 680gtx to see if it makes any difference by each test? I see you used 306.97, thanks