fh50

GPU Score

Recommended Posts

I have a I7 4770 and the CPU scores seem to be within range (821) but my question is about the GPU scores.I have seen numbers that are in the 800 range with the internal GPU but mine is 130. I am running Windows 8 pro with 16 GB of 1866mhz memory on an ASUS Z87 Expert MB. Any suggestions on why the diparity?

Share this post


Link to post
Share on other sites
NovaBench Score: 2894

10/1/2013 8:07:04 PM
Microsoft Windows 7 Ultimate
Intel Core i74930K 3.40GHz @ 3401 MHz
Graphics Card: NVIDIA GeForce GTX 780

16325 MB System RAM (Score: 266)
- RAM Speed: 17699 MB/s

CPU Tests (Score: 1203)
- Floating Point Operations/Second: 308545368
- Integer Operations/Second: 1692464232
- MD5 Hashes Generated/Second: 1608622

Graphics Tests (Score: 1391)
- 3D Frames Per Second: 3512

Hardware Tests (Score: 34)
- Primary Partition Capacity: 223 GB
- Drive Write Speed: 169 MB/s

Lovin it!

Share this post


Link to post
Share on other sites

Hi,

the graphic score of my Radeon HD 5750 changed from 364 to 291. Is it possible ?

Now i use Windows 8, bevore I used Windows 7, now i used catelyst 13.10, before the 12.10 version.

What do you mean ?

regards from Arno

Share this post


Link to post
Share on other sites

I have a I7 4770 and the CPU scores seem to be within range (821) but my question is about the GPU scores.I have seen numbers that are in the 800 range with the internal GPU but mine is 130. I am running Windows 8 pro with 16 GB of 1866mhz memory on an ASUS Z87 Expert MB. Any suggestions on why the diparity?

Can you post your Novabench score?

Hi,

the graphic score of my Radeon HD 5750 changed from 364 to 291. Is it possible ?

Now i use Windows 8, bevore I used Windows 7, now i used catelyst 13.10, before the 12.10 version.

What do you mean ?

regards from Arno

13.10 is the beta driver. Try 13.9

Share this post


Link to post
Share on other sites

Verified NovaBench Score: 1275


Test run on October 3, 2013 at 21:23
Microsoft Windows 8 Pro

intel.png Intel Core i7-4770 3.40GHz running at 3455 MHz

Intel HD Graphics 4600 GPU

16065 MB System RAM (Score: 247)
- RAM Speed: 14182 MB/s

CPU Tests (Score: 857)
- Floating Point Operations/Second: 206467408
- Integer Operations/Second: 1028810560
- MD5 Hashes Generated/Second: 1428976

Graphics Tests (Score: 129)
- 3D Frames Per Second: 395

Hardware Tests (Score: 42)
- Primary Partition Capacity: 251 GB
- Drive Write Speed: 237 MB/s

Share this post


Link to post
Share on other sites

There is a bug affecting some computers running Novabench where scores for dedicated graphics are attributed to Intel graphics. The 129 score is probably correct for the 4600, not 800 as it may say in the database. I will clean up the database when this bug is fixed.

Share this post


Link to post
Share on other sites

Hi,

for my Radeon HD5750 there was a nova bench grafic-score of 300- (Windows 8, Direct x 11 , Samsung SSD).

Now i used this Grafik Card for another PC with similar Hardware.

Only a HDD not a SSD and with Windos XP and directx 9.3.

In Novabench the same grafikcard has only a Novabench grafik-score of 30 !

What could be happend, that the grafic-card losses 90 % of grafic score?

Share this post


Link to post
Share on other sites

It may not be powered properly in the other machine? Make sure the additional power is plugged in.

Also make sure the drivers are properly installed.

Share this post


Link to post
Share on other sites

A majority of modern laptops are also affected by Novabench reporting the main video output as being the IGP, which is understandable since it's connected through LVDS and not checking for a second display adapter. From my understanding, seems both Nvidia Optimus and AMD Switchable Graphics suffer from this issue.

I'm rather puzzled at how the database lists Geforce GT 555M numbers so low (average of 81), given that I've been getting an average of 240.

EDIT: Does Hardware Tests score take into account Primary Partition Capacity that much? If it does, it shouldn't weigh nearly as much when compared to Drive Write Speed, in all honesty.

512393.png

Edited by tribaljet

Share this post


Link to post
Share on other sites

Fixing the GPU listing is top of the list, for sure.

The hardware test no longer includes the drive size in the score at all.

Share this post


Link to post
Share on other sites

I've noticed what I think is a bug. After running Novabench on an older laptop, I've saved the results to a file, then I've opened them on the laptop of the banner of my previous post, but it seems only results were correct. OS, CPU and GPU were all from the Core i7 system instead of the Core Duo system where Novabench ran.

EDIT: I've also noticed that besides HDD size being more important than write speeds for scores, it seems RAM amounts also matter more than speeds. A clear example of this is as follows:

http://novabench.com/view/513101 - My score

http://novabench.com/view/514216 - System with more but slower memory

As it can be seen, my RAM is over 3GB/s faster than the other system, yet that system has a higher RAM score due to having 16GB of RAM as opposed to mine that has 8GB. I honestly believe that speeds should be the only metric to count for scores, but if RAM/HDD size were to still be taken into account, at least it should weigh significantly less.

Edited by tribaljet

Share this post


Link to post
Share on other sites

Thanks for looking into the matter.

Also, out of curiosity, but are you set in terms of users with different systems for testing betas?

513101.png

Yes, beta is closed at the moment. It might reopen in the future.

I've noticed what I think is a bug. After running Novabench on an older laptop, I've saved the results to a file, then I've opened them on the laptop of the banner of my previous post, but it seems only results were correct. OS, CPU and GPU were all from the Core i7 system instead of the Core Duo system where Novabench ran.

EDIT: I've also noticed that besides HDD size being more important than write speeds for scores, it seems RAM amounts also matter more than speeds. A clear example of this is as follows:

http://novabench.com/view/513101 - My score

http://novabench.com/view/514216 - System with more but slower memory

As it can be seen, my RAM is over 3GB/s faster than the other system, yet that system has a higher RAM score due to having 16GB of RAM as opposed to mine that has 8GB. I honestly believe that speeds should be the only metric to count for scores, but if RAM/HDD size were to still be taken into account, at least it should weigh significantly less.

That could be a bug. The next version was written from scratch, so it shouldn't be an issue when it comes out.

As for the RAM, the amount does factor in, but the increase it makes to the score tapers off quickly. The reason for this is, while RAM speed is very important, for typical use the difference between 4GB and 8GB of RAM for example can mean considerably less paging to the disk, which can speed things up a lot.

Share this post


Link to post
Share on other sites

Good point about potencial paging issues, but if RAM amount score scales evenly, it's slightly unfair, given that systems with 8GB+ aren't really affected by paging issues, therefore it might perhaps make sense that after x amount of RAM, there is less of a score bonus.

Can more accurate system timers, such as enabling HPET, affect Novabench's results in any meaningful way?

Also, is Novabench's new version going to be released in an installer form or will binaries also be made available?

Share this post


Link to post
Share on other sites

Good point about potencial paging issues, but if RAM amount score scales evenly, it's slightly unfair, given that systems with 8GB+ aren't really affected by paging issues, therefore it might perhaps make sense that after x amount of RAM, there is less of a score bonus.

Yeah, the RAM score bonus does go down as RAM increases. I can't remember off-hand what the formula is though.

Can more accurate system timers, such as enabling HPET, affect Novabench's results in any meaningful way?

Possibly, but I haven't done any experiments with this. The specific timer(s) being used aren't known to Novabench.

Also, is Novabench's new version going to be released in an installer form or will binaries also be made available?

At this point it looks like it will need an installer due to certain DirectX dependencies we need to check for and install. This is the same reason the current version needs an installer. Due to Microsoft's licensing we can't just drop the required Direct3D DLLs into a zip along with Novabench.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now