StarBound's 5870 Crossfire Journal
-
- Registered Pervert
- Posts: 6879
- Joined: 30 Jul 2004, 02:00
- Processor: Intel i7 4790k
- Motherboard: MSI Z97 Gaming 7
- Graphics card: MSI GTX780Ti Gaming
- Memory: G.Skill Sniper 1866mhz 16GB
- Location: The Greater Unknown
- Contact:
StarBound's 5870 Crossfire Journal
I decided while I could spam the AMD vs Nvidia thread I would rather make a new one with all my findings.
Installation:
I had a hard time installing the cards due to connectors being in the way, a pci-e 1x slot that didnt work due to ACS-6xxx driver screwing everything up, and that I had to move my case every time I needed to insert and swap cards around.
Finally I managed to get everything working and try out crossfire for myself.
Figures doesn't lie:
They sure don't. Here are the scores from the unigine benchmarks.
Tropics Demo v1.3
FPS: 72.6
Scores: 1829
Min FPS: 52.5
Max FPS: 115.2
---crossfire---
FPS: 130.7
Scores: 3293
Min FPS: 47.0
Max FPS: 219.1
Heaven Benchmark v2.1
FPS: 35.5
Scores: 894
Min FPS: 13.7
Max FPS: 93.2
---crossfire---
FPS: 64.8
Scores: 1632
Min FPS: 3.5
Max FPS: 178.6
---crossfire extreme tessellation---
FPS: 40.3
Scores: 1014
Min FPS: 12.5
Max FPS: 144.3
As you can see it is a 90% increase in the maximum and average framerate. For unknown reasons the lowest framerate is also lower with crossfire.
Can you see the difference?:
To be perfectly honest if it was not for the benchmark numbers I would only have seen the one place that made a difference and that would be the start of the tropics demo. Unfortunately the crysis benchmark didnt file it but it was around 35-40fps without crossfire and 40fps with crossfire on the flyby. The cpu demos which I prefer testing remains as stubborn as ever with cpu2 not being able to get past 20fps and slowed down to 5 fps. Metro 2033 demo does not come with a benchmark tool but I felt that a single card performed better. Again it's what I saw and didn't have the figures to back it up or prove it wrong. Just Cause 2 benchmark capped at 60fps on a single card so I was not bothered to even run a crossfire test.
Vsync. Cripple for quality.:
My screen refreshes at 60hz. I try to play all my games with vsync. This means my max number of frames per second can only be 60 which is fine by me. Unless I get a 120hz screen or go 3D I have no other reason for getting it higher than that. Solid pictures at the end of the day is what I want.
What's that sound? ...and smell?:
Crossfiring 2 5870s is hot and loud. A single 5870 isnt too noisy under load and only clocks at 70C max if it even gets that high. With crossfire my cards were running at near 80C and I could factor in the fans as an annoyance.
Out of memory!?:
With 4gb ram my philosify is "Use the !@#$ing RAM and leave the page file alone!". But sadly windows has another idea. While playing the Metro demo with a 256mb page file and in crossfire the game just closed and windows gave me a low on memory warning. The game didnt even exceed the 2.8gb mark on my resource manager. Starcraft 2 had the exact same issue. No where near 4gb used the game closed, low memory notification, 1.5gb away from 4gb. This does not happen if your page file is the same size as your ram or if you let windows manage your page file. I do not like or trust windows with such a task so I force it with a small page file to use that extra nano second responce times that I pay so much for.
Final verdict:
Crossfire works. There is an increase in performance. To me it isn't worth it. The issues I am having might be restricted to me but when I bought the first card I didnt want fermi noise or heat and crossfire is working well in getting it on par with the 470/480gtx. With most games running at 60fps already I dont see the point in getting extra speed. Future games will need better cards. Future cards will have features that this generation doesnt have. This only strengthens it that if you want to go sli or crossfire you do it immediately and dont regret what you got if for. Otherwise a single card solution is the better idea.
Installation:
I had a hard time installing the cards due to connectors being in the way, a pci-e 1x slot that didnt work due to ACS-6xxx driver screwing everything up, and that I had to move my case every time I needed to insert and swap cards around.
Finally I managed to get everything working and try out crossfire for myself.
Figures doesn't lie:
They sure don't. Here are the scores from the unigine benchmarks.
Tropics Demo v1.3
FPS: 72.6
Scores: 1829
Min FPS: 52.5
Max FPS: 115.2
---crossfire---
FPS: 130.7
Scores: 3293
Min FPS: 47.0
Max FPS: 219.1
Heaven Benchmark v2.1
FPS: 35.5
Scores: 894
Min FPS: 13.7
Max FPS: 93.2
---crossfire---
FPS: 64.8
Scores: 1632
Min FPS: 3.5
Max FPS: 178.6
---crossfire extreme tessellation---
FPS: 40.3
Scores: 1014
Min FPS: 12.5
Max FPS: 144.3
As you can see it is a 90% increase in the maximum and average framerate. For unknown reasons the lowest framerate is also lower with crossfire.
Can you see the difference?:
To be perfectly honest if it was not for the benchmark numbers I would only have seen the one place that made a difference and that would be the start of the tropics demo. Unfortunately the crysis benchmark didnt file it but it was around 35-40fps without crossfire and 40fps with crossfire on the flyby. The cpu demos which I prefer testing remains as stubborn as ever with cpu2 not being able to get past 20fps and slowed down to 5 fps. Metro 2033 demo does not come with a benchmark tool but I felt that a single card performed better. Again it's what I saw and didn't have the figures to back it up or prove it wrong. Just Cause 2 benchmark capped at 60fps on a single card so I was not bothered to even run a crossfire test.
Vsync. Cripple for quality.:
My screen refreshes at 60hz. I try to play all my games with vsync. This means my max number of frames per second can only be 60 which is fine by me. Unless I get a 120hz screen or go 3D I have no other reason for getting it higher than that. Solid pictures at the end of the day is what I want.
What's that sound? ...and smell?:
Crossfiring 2 5870s is hot and loud. A single 5870 isnt too noisy under load and only clocks at 70C max if it even gets that high. With crossfire my cards were running at near 80C and I could factor in the fans as an annoyance.
Out of memory!?:
With 4gb ram my philosify is "Use the !@#$ing RAM and leave the page file alone!". But sadly windows has another idea. While playing the Metro demo with a 256mb page file and in crossfire the game just closed and windows gave me a low on memory warning. The game didnt even exceed the 2.8gb mark on my resource manager. Starcraft 2 had the exact same issue. No where near 4gb used the game closed, low memory notification, 1.5gb away from 4gb. This does not happen if your page file is the same size as your ram or if you let windows manage your page file. I do not like or trust windows with such a task so I force it with a small page file to use that extra nano second responce times that I pay so much for.
Final verdict:
Crossfire works. There is an increase in performance. To me it isn't worth it. The issues I am having might be restricted to me but when I bought the first card I didnt want fermi noise or heat and crossfire is working well in getting it on par with the 470/480gtx. With most games running at 60fps already I dont see the point in getting extra speed. Future games will need better cards. Future cards will have features that this generation doesnt have. This only strengthens it that if you want to go sli or crossfire you do it immediately and dont regret what you got if for. Otherwise a single card solution is the better idea.
Re: StarBound's 5870 Crossfire Journal
No offence, but obviously this isn't working out for you. The problem is not with Windows, its with you.I do not like or trust windows with such a task so I force it with a small page file to use that extra nano second responce times that I pay so much for.
As I understand it, you will only have slow performance when part of the loaded programs that are in use at the moment is forced to operate from within the page file. That will not easily happen with 4GB I don't think, but only if your page file is big enough. I think there might be some confusion as to what the page file is there for (maybe from my side), which is not as simple as it acting like extra RAM when RAM space runs out AFAIK. AFAIK Windows still uses the page file, even if you have 16GB of RAM. What I am saying is, if Windows complains about running out of memory, then maybe you should let it manage it automatically after all.
Humming tunes while you play might help with that.What's that sound?
Vicks. Apply it.and smell?
If you get rid of your second card, why not consider going for a good NVidia card (GTX460 or such) so you can run PhysX at full tilt? At least then you'll gain something tangible from getting a second card.Can you see the difference?:
"It is the mark of an educated mind to be able to entertain a thought without accepting it." - Aristotle
Intel i5 2500; AsRock Z77 Extreme 4; Asus GTX580; 4x 2GB DDR3 1333; Intel 520 240GB SSD + 2x WD 3TB + 2TB Samsung; Samsung 22X DVD/RW; 23" LG W2343T-PF; Huntkey 700W
Intel i5 2500; AsRock Z77 Extreme 4; Asus GTX580; 4x 2GB DDR3 1333; Intel 520 240GB SSD + 2x WD 3TB + 2TB Samsung; Samsung 22X DVD/RW; 23" LG W2343T-PF; Huntkey 700W
-
- Registered Pervert
- Posts: 6879
- Joined: 30 Jul 2004, 02:00
- Processor: Intel i7 4790k
- Motherboard: MSI Z97 Gaming 7
- Graphics card: MSI GTX780Ti Gaming
- Memory: G.Skill Sniper 1866mhz 16GB
- Location: The Greater Unknown
- Contact:
Re: StarBound's 5870 Crossfire Journal
A single card is fast enough. And it comes back to the reason behind going ATI in the first place. Less noise, less heat, less power consumption. I knew crossfire would render the last one useless but as far as noise and heat goes it isnt what I want.
Like I said crossfire is worth it for anyone that wants it if they dont have the gripes I have. It just doesnt do it for me.
Like I said crossfire is worth it for anyone that wants it if they dont have the gripes I have. It just doesnt do it for me.
Re: StarBound's 5870 Crossfire Journal
I like those results!
I have 2GB of RAM. I set my windows page file to 2MB (the smallest size.) I set it to Off in Windows XP, but Windows promptly turned it back on and set it to Auto Managed on reboot. When I switched to Win7 I set it to 2MB again but found myself running short of memory so I set it to 5120MB.
Are you running a 32bit or 64bit version of Win7? And which edition of Win7 are you running?
I have 2GB of RAM. I set my windows page file to 2MB (the smallest size.) I set it to Off in Windows XP, but Windows promptly turned it back on and set it to Auto Managed on reboot. When I switched to Win7 I set it to 2MB again but found myself running short of memory so I set it to 5120MB.
Are you running a 32bit or 64bit version of Win7? And which edition of Win7 are you running?
Important Thread:
- Hey everybody, I have returned (2013) and I am not Dead.
- Explaining RAM - SDRAM, DDRx and GDDR
-
- Registered User
- Posts: 3515
- Joined: 28 Feb 2004, 02:00
- Processor: Xeon E5620
- Motherboard: Asus P6T6 Workstation
- Graphics card: MSI GTX770
- Memory: 24GB Hynix
- Location: ::1
Re: StarBound's 5870 Crossfire Journal
Interesting. In XP64 I can set the page file to 0MB and it will stay like that. I also disabled the page file totally, work pretty well IMO. 4GB is getting a bit to small though. My next machine I'm planning for next year sometime will have 24GB, that should be enough not to need a pagefile for the next few years.
Windows memory management is in anyway crap compared to Linux. Linux will use as much RAM as possible, to minimize drive access.
Windows memory management is in anyway crap compared to Linux. Linux will use as much RAM as possible, to minimize drive access.
-
- Registered Pervert
- Posts: 6879
- Joined: 30 Jul 2004, 02:00
- Processor: Intel i7 4790k
- Motherboard: MSI Z97 Gaming 7
- Graphics card: MSI GTX780Ti Gaming
- Memory: G.Skill Sniper 1866mhz 16GB
- Location: The Greater Unknown
- Contact:
Re: StarBound's 5870 Crossfire Journal
64-bit.
All I know is some programs complain if I dont have it on. I originally had it off. Then starcraft 2 complained about the 1gb free ram that isnt enough. Then I enabled 256mb pagefile which fixed it. And last night with crossfire starcraft 2 simply closed due to not enough memory. I dont pretend to know why windows needs a pagefile but if I could I would take 1gb ram and use it as a pagefile.
All I know is some programs complain if I dont have it on. I originally had it off. Then starcraft 2 complained about the 1gb free ram that isnt enough. Then I enabled 256mb pagefile which fixed it. And last night with crossfire starcraft 2 simply closed due to not enough memory. I dont pretend to know why windows needs a pagefile but if I could I would take 1gb ram and use it as a pagefile.
Re: StarBound's 5870 Crossfire Journal
You could always stick a 2GB or so USB stick into your machine and try to force Windows to use that as the drive for your swap file. You could also use it as a ReadyBoost drive.
Alternatively, and I think that this is an avenue worth pursuing, set up a 512MB or 1024MB RAM drive and put your swap file on that. Please note that I haven't looked into this yet, so it might not even be possible.
Edit:
Alternatively, and I think that this is an avenue worth pursuing, set up a 512MB or 1024MB RAM drive and put your swap file on that. Please note that I haven't looked into this yet, so it might not even be possible.
Edit:
Err, sorry, I missed this. What you are describing is a RAM drive.StarBound wrote:I dont pretend to know why windows needs a pagefile but if I could I would take 1gb ram and use it as a pagefile.
Important Thread:
- Hey everybody, I have returned (2013) and I am not Dead.
- Explaining RAM - SDRAM, DDRx and GDDR
-
- Registered Pervert
- Posts: 6879
- Joined: 30 Jul 2004, 02:00
- Processor: Intel i7 4790k
- Motherboard: MSI Z97 Gaming 7
- Graphics card: MSI GTX780Ti Gaming
- Memory: G.Skill Sniper 1866mhz 16GB
- Location: The Greater Unknown
- Contact:
Re: StarBound's 5870 Crossfire Journal
Isnt that essencially what a page file is? Swapping ram content to a file on your harddrive when its not in use?
Re: StarBound's 5870 Crossfire Journal
Thanks for the review SB.
There are 10 types of people in this world.
Those who understand binary and those who do not.
Those who understand binary and those who do not.
Re: StarBound's 5870 Crossfire Journal
It depends on the OS StarBound, specifically the algorithm that governs the virtual memory. In Windows XP it seems that the OS tried to keep as much physical memory free as possible, swapping stuff out to virtual memory (the swap file) as soon as possible. Vista, on the other hand, freaked everyone - including myself - out when it was first released because it seems like it grabs great swaths of physical memory and reserves it for it's own uses, even when it actually needs much less, e.g. when I installed it on my computer a while back it only REALLY needed about 300MB but it had reserved over 700MB(!) of physical memory. Based on personal experience, it seems that the guys at MS have improved the algorithm for Win7 substantially.
The reason that a small RAM drive might work nicely is that data isn't being moved to and from a very slow HDD, it's being "moved" to another location in memory. Since Windows thinks that you have a swap file, programs that require a swap file now should work ok.
The reason that a small RAM drive might work nicely is that data isn't being moved to and from a very slow HDD, it's being "moved" to another location in memory. Since Windows thinks that you have a swap file, programs that require a swap file now should work ok.
Important Thread:
- Hey everybody, I have returned (2013) and I am not Dead.
- Explaining RAM - SDRAM, DDRx and GDDR
Re: StarBound's 5870 Crossfire Journal
MAN that was a long typing session on my cellphone!!!
Important Thread:
- Hey everybody, I have returned (2013) and I am not Dead.
- Explaining RAM - SDRAM, DDRx and GDDR
Re: StarBound's 5870 Crossfire Journal
Although in SB's case it isn't a slow HDD since he has a solid state drive.GDI_Lord wrote:The reason that a small RAM drive might work nicely is that data isn't being moved to and from a very slow HDD, it's being "moved" to another location in memory.
There are 10 types of people in this world.
Those who understand binary and those who do not.
Those who understand binary and those who do not.
-
- Registered Pervert
- Posts: 6879
- Joined: 30 Jul 2004, 02:00
- Processor: Intel i7 4790k
- Motherboard: MSI Z97 Gaming 7
- Graphics card: MSI GTX780Ti Gaming
- Memory: G.Skill Sniper 1866mhz 16GB
- Location: The Greater Unknown
- Contact:
Re: StarBound's 5870 Crossfire Journal
SSD still reads and writes in milli seconds. RAM works in nano.
I worked out that if I keep writing and reading my SSD should die out in 9 years at 365/24/7 usage. But I wont exactly put the tech to test It's too expensive and new to try it. Maybe when it finally dies I will have a good idea.
I worked out that if I keep writing and reading my SSD should die out in 9 years at 365/24/7 usage. But I wont exactly put the tech to test It's too expensive and new to try it. Maybe when it finally dies I will have a good idea.
Re: StarBound's 5870 Crossfire Journal
Compared to moving data to and from RAM, even an SSD is like watching a movie of a tranquilised sloth waking up in super-slow motion.
Important Thread:
- Hey everybody, I have returned (2013) and I am not Dead.
- Explaining RAM - SDRAM, DDRx and GDDR
-
- Registered Pervert
- Posts: 6879
- Joined: 30 Jul 2004, 02:00
- Processor: Intel i7 4790k
- Motherboard: MSI Z97 Gaming 7
- Graphics card: MSI GTX780Ti Gaming
- Memory: G.Skill Sniper 1866mhz 16GB
- Location: The Greater Unknown
- Contact:
Re: StarBound's 5870 Crossfire Journal
"Sloth! Stuck in bullit time since birth!"
-
- Registered Pervert
- Posts: 6879
- Joined: 30 Jul 2004, 02:00
- Processor: Intel i7 4790k
- Motherboard: MSI Z97 Gaming 7
- Graphics card: MSI GTX780Ti Gaming
- Memory: G.Skill Sniper 1866mhz 16GB
- Location: The Greater Unknown
- Contact:
Re: StarBound's 5870 Crossfire Journal
My final crossfire report. Metro 2033 very high details:
Options: Resolution: 1920 x 1080; DirectX: DirectX 11; Quality: Very High; Antialiasing: AAA; Texture filtering: AF 16X; Advanced PhysX: Disabled; Tesselation: Enabled; DOF: Enabled
* Average Framerate: 24.00
* Max. Framerate: 51.43
* Min. Framerate: 8.40
Crossfire
Options: Resolution: 1920 x 1080; DirectX: DirectX 11; Quality: Very High; Antialiasing: AAA; Texture filtering: AF 16X; Advanced PhysX: Disabled; Tesselation: Enabled; DOF: Enabled
* Average Framerate: 36.33
* Max. Framerate: 177.27
* Min. Framerate: 5.80
The results are that you should clearly be able to see a difference when using 2 cards. The lower is lower, the average and max were higher.
Options: Resolution: 1920 x 1080; DirectX: DirectX 11; Quality: Very High; Antialiasing: AAA; Texture filtering: AF 16X; Advanced PhysX: Disabled; Tesselation: Enabled; DOF: Enabled
* Average Framerate: 24.00
* Max. Framerate: 51.43
* Min. Framerate: 8.40
Crossfire
Options: Resolution: 1920 x 1080; DirectX: DirectX 11; Quality: Very High; Antialiasing: AAA; Texture filtering: AF 16X; Advanced PhysX: Disabled; Tesselation: Enabled; DOF: Enabled
* Average Framerate: 36.33
* Max. Framerate: 177.27
* Min. Framerate: 5.80
The results are that you should clearly be able to see a difference when using 2 cards. The lower is lower, the average and max were higher.
-
- Registered User
- Posts: 673
- Joined: 11 Aug 2010, 15:32
- Location: Durban, Pinetown
- Contact:
Re: StarBound's 5870 Crossfire Journal
I'm glad to see I'm not the only person who has to use my phone to view this forum!GDI_Lord wrote:MAN that was a long typing session on my cellphone!!!
And gprs is such a pain!
[i7 2600K @ 4.6 ghz 1.35V] [Asus P8P67 Deluxe] [G-Skill RipjawsX 2x2GB DDR3 1600 CL6 (6-8-6-24) 1.5V]
[Sapphire Radeon 6870 @ 975/1200 1.2V] [Coolermaster V6 GT] [WD Caviar Black 640GB SATA3 6Gbps]
[Samsung SA350 23" LED] [NZXT Phantom White] [Corsair TX750W V2] [Zalman RS6F-USB 5.1 Surround Sound Headphones] My rig and benchmarks
[Sapphire Radeon 6870 @ 975/1200 1.2V] [Coolermaster V6 GT] [WD Caviar Black 640GB SATA3 6Gbps]
[Samsung SA350 23" LED] [NZXT Phantom White] [Corsair TX750W V2] [Zalman RS6F-USB 5.1 Surround Sound Headphones] My rig and benchmarks
-
- Registered User
- Posts: 673
- Joined: 11 Aug 2010, 15:32
- Location: Durban, Pinetown
- Contact:
Re: StarBound's 5870 Crossfire Journal
I suppose it works looking at the averages. But I'm not impressed with xfire having the lowest fps
[i7 2600K @ 4.6 ghz 1.35V] [Asus P8P67 Deluxe] [G-Skill RipjawsX 2x2GB DDR3 1600 CL6 (6-8-6-24) 1.5V]
[Sapphire Radeon 6870 @ 975/1200 1.2V] [Coolermaster V6 GT] [WD Caviar Black 640GB SATA3 6Gbps]
[Samsung SA350 23" LED] [NZXT Phantom White] [Corsair TX750W V2] [Zalman RS6F-USB 5.1 Surround Sound Headphones] My rig and benchmarks
[Sapphire Radeon 6870 @ 975/1200 1.2V] [Coolermaster V6 GT] [WD Caviar Black 640GB SATA3 6Gbps]
[Samsung SA350 23" LED] [NZXT Phantom White] [Corsair TX750W V2] [Zalman RS6F-USB 5.1 Surround Sound Headphones] My rig and benchmarks
Re: StarBound's 5870 Crossfire Journal
How timeous! An article that examines RAM disks over at Tom's Hardware.
Important Thread:
- Hey everybody, I have returned (2013) and I am not Dead.
- Explaining RAM - SDRAM, DDRx and GDDR
-
- Registered Pervert
- Posts: 6879
- Joined: 30 Jul 2004, 02:00
- Processor: Intel i7 4790k
- Motherboard: MSI Z97 Gaming 7
- Graphics card: MSI GTX780Ti Gaming
- Memory: G.Skill Sniper 1866mhz 16GB
- Location: The Greater Unknown
- Contact:
Re: StarBound's 5870 Crossfire Journal
Haha. Posted a thread about it before your reply XD
But yes a weird string of co-incidents.
But yes a weird string of co-incidents.