Best silent 9600GT!

They make noise, too.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Sun Apr 20, 2008 6:31 am

HammerSandwich wrote:I bought this card, because I'm lazy & the price was right. Nice card, no complaints.
Why does that card have what looks like 2 molex connectors?

HammerSandwich
*Lifetime Patron*
Posts: 1288
Joined: Sat Oct 25, 2003 3:21 pm
Location: 15143, USA
Contact:

Post by HammerSandwich » Sun Apr 20, 2008 7:15 am

Not sure what you're referring to.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Sun Apr 20, 2008 7:49 am

HammerSandwich wrote:Not sure what you're referring to.
Click on the picture of that card you linked to and look at the power connector:

http://www.newegg.com/Product/Product.a ... 6814134040

bonestonne
Posts: 1839
Joined: Mon Feb 05, 2007 2:10 pm
Location: Northern New Jersey
Contact:

Post by bonestonne » Sun Apr 20, 2008 7:59 am

JVM wrote:What Ninja model is that? Another question I have is if you need to replace your rear Nexus, how are you going to get those rubber mounts into the fan? Looks like you will have to take off the heatsink, or?

Just thought of two more questions, how do you like that Gigabyte board ( EP35-DS4?)? And, how are you reading fan speeds and temps?
tehcrazybob has a Ninja Rev B, with a bolt-thru kit.

The two molex connectors that you see go into a single 6 pin power connector for the 9600, in case the power supply you have doesn't have a PCI-e power connector.
--------------
Will try testing loads soon...have to find the software i had before.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Sun Apr 20, 2008 8:18 am

bonestonne wrote:
JVM wrote:What Ninja model is that? Another question I have is if you need to replace your rear Nexus, how are you going to get those rubber mounts into the fan? Looks like you will have to take off the heatsink, or?

Just thought of two more questions, how do you like that Gigabyte board ( EP35-DS4?)? And, how are you reading fan speeds and temps?
tehcrazybob has a Ninja Rev B, with a bolt-thru kit.

The two molex connectors that you see go into a single 6 pin power connector for the 9600, in case the power supply you have doesn't have a PCI-e power connector.
--------------
Will try testing loads soon...have to find the software i had before.
I don't understand that about the power connector. There is one end of that connector that should be a 6-pin connection to the video card, and the other end has what looks like 2 molex connectors. If you connect those 2 molex-type connectors to the video card, then what would you do with the other end? Also, if you have a PSU with the 6-pin connection, then how would you use that power cable?

I like the idea of the Bolt-Thru kit, but how is it done? Does it require some sort of modification to the board?

bonestonne
Posts: 1839
Joined: Mon Feb 05, 2007 2:10 pm
Location: Northern New Jersey
Contact:

Post by bonestonne » Sun Apr 20, 2008 8:21 am

no, the bolt-thru kit doesn't need any mods to the board, and is made by Thermalright. So conveniently named: Thermalright LGA775 Bolt-thru kit.

The power adapter for the 9600GT, you're just thinking about it backwards. you plug the molex connectors to the power supply, and the 6 pin into the card. i have to use the adapter for my card because my PSU doesn't have the 6 pin cable, and its flawless. no problems yet, and i'm about to hit it with ATi-Tools.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Sun Apr 20, 2008 8:29 am

bonestonne wrote:no, the bolt-thru kit doesn't need any mods to the board, and is made by Thermalright. So conveniently named: Thermalright LGA775 Bolt-thru kit.

The power adapter for the 9600GT, you're just thinking about it backwards. you plug the molex connectors to the power supply, and the 6 pin into the card. i have to use the adapter for my card because my PSU doesn't have the 6 pin cable, and its flawless. no problems yet, and i'm about to hit it with ATi-Tools.
I know the 6-pin connector goes to the board. So, the picture that shows what's labeled Power Cable is actually a Power Adapter for those without a 6-pin connection. Newegg should have labeled that Power Adapter!

Regarding the Bolt-Thru kit, you use that instead of the push-pin thing that comes with the Ninja and doesn't require any modification at all?

bonestonne
Posts: 1839
Joined: Mon Feb 05, 2007 2:10 pm
Location: Northern New Jersey
Contact:

Post by bonestonne » Sun Apr 20, 2008 9:00 am

all you have to do with the Ninja is get the push-pins off. There's a small black tab on each pushpin that you bend back carefully (as to not break it) and you can then slide the clear plastic piece out. Once that is out, you can carefully pinch the end of the black clip, and push it out of the base. with the bolt-thru kit, you just use the spring-loaded screws with the standard holes the pushpins used to be in, and you also need to use the backplate supplied with it. you do have to remove the motherboard to use the bolt-thru kit.

no other mods to the ninja.
-------------------------------------------
Okay! I have load results from my 9600GT!

Code: Select all

Idle:
Case Open: 35C it took nearly 10 minutes with the case open for it to settle
Case Closed: 38C pretty much the same temp as it was when i did a cold boot after a few minutes.

Load:
Case Open: 48C it took about 7 minutes for it to reach this temp, but it wasn't bouncing.
Case Closed: 52C for a while i thought it would have stayed at 51, but 8 minutes in, it jumped up to 52 and sat there.
Tested with ATi-Tool 0.26.

Also, with the 3D view on, I was getting 787FPS with the case open, roughly 700 with the case closed.

The CPU hadn't changed drastically while I was testing, but i checked it anyway:

Code: Select all

Idle:
Case Open: 23C
Case Closed: 25C

ATi-Tools:
Case Open: 33C
Case Closed: 36C
Temps for the CPU taken with Speedfan 4.33

One final note is that when ATi-Tools was running, it only took about 5 minutes for the PSU to really ramp up and start blowing out some warm air.

Ambient room temp right now and during the test is: 21.6C

tehcrazybob
Friend of SPCR
Posts: 356
Joined: Wed Jan 16, 2008 8:56 pm
Location: Council Bluffs, Iowa
Contact:

Post by tehcrazybob » Sun Apr 20, 2008 10:48 am

JVM wrote:What Ninja model is that? Another question I have is if you need to replace your rear Nexus, how are you going to get those rubber mounts into the fan? Looks like you will have to take off the heatsink, or?

Just thought of two more questions, how do you like that Gigabyte board ( EP35-DS4?)? And, how are you reading fan speeds and temps?
As Bonestonne noted, it's a Ninja Rev B and a Thermalright bolt-through kit. And you're right, removing the Nexus could be a problem. I'm hoping I could just leave the rubber mounts attached to the fan rather than the case, and feed them backwards through the holes in the case. It might take tweezers and patience, but at least I wouldn't have to pull the motherboard.

I'd definitely recommend the bolt-through kit, as it seems a bit sturdier than the push-pins. You just have to remove the plastic pins from the mount that comes with the Ninja, and thread the Thermalright bolts through those holes. I don't have a good picture of it, but here's a close-up from one of the pictures I posted earlier:
Image

As for the board, it's actually a P35-DS3L. I absolutely love it and would recommend it to anyone, but if you're concerned with power savings I understand the EP35- series is slightly better. I'm not sure what the difference is between the DS3L and DS4, though.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Sun Apr 20, 2008 11:12 am

How many fan headers have rpm monitoring? Also, are there any issues connecting to the CPU or System 1 fan header that are either under or very close to the heatsink? --I'm going by the EP35-DS4 board.

I think your board is 1.0 or 1.1 for PCI-E 16 as opposed to 2.0, and I think you have the 9600 GT which is 2.0. Have you thought about what the performance difference is between 1.0 and 2.0?

tehcrazybob
Friend of SPCR
Posts: 356
Joined: Wed Jan 16, 2008 8:56 pm
Location: Council Bluffs, Iowa
Contact:

Post by tehcrazybob » Sun Apr 20, 2008 1:32 pm

JVM wrote:How many fan headers have rpm monitoring? Also, are there any issues connecting to the CPU or System 1 fan header that are either under or very close to the heatsink? --I'm going by the EP35-DS4 board.

I think your board is 1.0 or 1.1 for PCI-E 16 as opposed to 2.0, and I think you have the 9600 GT which is 2.0. Have you thought about what the performance difference is between 1.0 and 2.0?
All four fan headers offer RPM monitoring, and I believe two of them allow speed control, both by voltage adjust for 3-pin fans and PWM for 4-pin fans. None of the fan headers are directly under the Ninja, so while it might be a bit difficult to reach some, they're all usable.

As for PCIe generations, you are correct. The DS3L is only version 1.1, but it's not much of an issue. Pretty much the only cards currently on the market which need the extra bandwidth of 2.0 are the dual-GPU cards; 9800GX2 and HD3870X2. For a 9600gt there's absolutely no performance drop due to using the older connection.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Sun Apr 20, 2008 2:21 pm

Well, that is good news about the 2.0 issue since I plan on building with the 9600 GT, probably that ECS passive card that looks good.

What I would like to know is how does it work if connecting 3-pin fans to all those headers including the 4-pin headers, and if you can monitor the rpm speed with Speedfan or other outside utility?

I wouldn't want the motherboard controlling my fans and prefer to handle it as I do now. The software program included with the board (Easy Tune?) probably has to be started in Windows and doesn't display unless maximized, and I'm guessing about that.

I took a look on Newegg and saw an X38 board, the EX38-DS4, that is priced nicely for an X38 board. I know it isn't absolutely necessary, but I like the idea of having 2.0 available for possible future use; on the other hand, the price of your board is absolutely great! I haven't truly decided on where I'm going yet. I did notice the EX38-DS4 has a small plate on the back, not a big plate like the their other X38 board, but it is there and of some concern mounting the Ninja. Pictures are here: http://www.newegg.com/Product/Product.a ... 6813128089

bonestonne
Posts: 1839
Joined: Mon Feb 05, 2007 2:10 pm
Location: Northern New Jersey
Contact:

Post by bonestonne » Sun Apr 20, 2008 3:11 pm

That's just a play for the northbridge chipset, and it should not come in contact with the backplate in the Thermalright LGA775 Bolt-thru kit. if there is, there should be no problem taking a hacksaw to the blackplate and, well, adjusting it to fit.

speedfan and other utilities work by addressing the hardware bus that the fan is plugged into, and changing the voltage to the fan. generally, you can find a compatibility list for the programs, which tell you what motherboards will work, and what wont. i prefer a hardware method, such as a zalman fanmate because it ensure that i will get my results instantly, i don't have to wait for a program to start.

i think that tehcrazybob is right, with the zero performance gain towards PCI express version 2, and in my honest opinion, for $200, its not worth the future proofing. look at everyone with the original Q6600...very soon after they got it, the G0 version hits the market...was it future proofing by getting it when it first came out? none of the kinks were worked out.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Sun Apr 20, 2008 3:59 pm

I was under the impression that 2.0 made a difference for cards with specs of 2.0.

As for the X38 board and the cost, you are probably right, especially after I read a review comparing both Gigabyte board (X35 & X38) with them being fairly equal overall in performance. I am not certain, but I think the X38 boards might run hotter than the P35 boards.

Seems like I likely will return to the P35 Gigabyte and save some bucks.

Oh, I am using Fanmate now and that is what I would use with the new build. I was asking about monitoring RPM speeds, not controlling the fans. Presently I use Hmonitor to see RPM speeds and temps. The thing is I don't know if Speedfan or Hmonitor can monitor fan speeds on the P35 Gigabyte board, which is why I asked tehcrazybob what he's using to monitor RPM speeds.

tehcrazybob
Friend of SPCR
Posts: 356
Joined: Wed Jan 16, 2008 8:56 pm
Location: Council Bluffs, Iowa
Contact:

Post by tehcrazybob » Sun Apr 20, 2008 8:38 pm

I'm measuring speeds and temperatures with SpeedFan, and it behaves as I mentioned in my last post - RPM monitoring on all headers as well as speed control on two.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Sun Apr 20, 2008 8:52 pm

tehcrazybob wrote:I'm measuring speeds and temperatures with SpeedFan, and it behaves as I mentioned in my last post - RPM monitoring on all headers as well as speed control on two.
Excellent! What about your temps? Also, what kind of RAM are you using?

And, are you using a front fan? Have you put a fan on the Accelero?

tehcrazybob
Friend of SPCR
Posts: 356
Joined: Wed Jan 16, 2008 8:56 pm
Location: Council Bluffs, Iowa
Contact:

Post by tehcrazybob » Mon Apr 21, 2008 7:01 am

At full load in a room at 23c, my idle/load temperatures are
CPU: 32/45°C
GPU: 45/65°C

The only fans in the system are the one in the Seasonic S12-II and the rear Nexus; the power supply never appears to ramp up from idle and the Nexus is running at around 900 RPM. The northbridge gets very warm to the touch (unsure the temp, as I can't find a sensor reading which makes sense for it), but no warmer than it did with the stock Intel cooler and the motherboard's automatic fan speed control, which is enabled by default.

The RAM is 2x1GB Transcend JetRAM, which was chosen by virtue of being 1) RAM and 2) cheap. It hasn't caused any problems.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Mon Apr 21, 2008 7:33 am

Your temps look all right to me, especially the 9600 GT without a fan on the Accelero running idle at 45C. My 7600 GT passive runs around 44/45C at idle.

Also very interesting is your CPU temp at idle being 32C. My AMD Athlon 64 3500+ Venice usually runs at 32C, except during summer, and I have the rear Scythe "E" rear case fan (recently switched from the "D" because of warmer weather) running around 940 RPM. I also have a front 92mm Nexus in my P150. And, my present heatsink/fan combo is Thermalright SI-120 with Scythe "D" at a bit over 800 RPM 12V.

I am curious as to your Chassis/Mainboard temperature, as that in my system is higher than the CPU temp.

Why haven't you disabled that Smart Fan thing in the BIOS?

Regarding your Chassis/Mainboard temperature, try Hmonitor--there is a free trial download from the Web site: http://www.hmonitor.net/

tehcrazybob
Friend of SPCR
Posts: 356
Joined: Wed Jan 16, 2008 8:56 pm
Location: Council Bluffs, Iowa
Contact:

Post by tehcrazybob » Mon Apr 21, 2008 8:18 am

I should mention that my processor's load state is undervolted to 1.056, which affects the load temperature. The idle state is unchanged, so the idle temperatures are the same as they would be ordinarily.

I might take a look at HMonitor, but I really don't think SpeedFan is missing any sensors. And I have no doubt that the northbridge is hotter than my CPU, as they generally are. However, chips like that are usually happy to 75 or 80c and safe to 100c, and I know mine's reasonable because I have no stability issues.

As for the fan speed, it is currently disabled. I had it enabled when I was using the stock cooler so the fan didn't run at 2000 RPM all the time. I mentioned it because with the stock top-down cooler, the northbridge was getting some secondary airflow off of it. However, because the fan spent most of its time at very low speed, the northbridge didn't get much airflow. Since Gigabyte felt their cooling was adequate in that situation, and the northbridge is currently no hotter than it was then, I believe it's completely safe.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Mon Apr 21, 2008 8:41 am

I don't use SpeedFan, but I am surprised you are having a problem reading your mainboard temp. Try Hmonitor and that should do it for you.

I should have also mentioned that I am quite impressed you accomplished all this without putting a fan on the Ninja. Of course, you could most likely reduce your mainboard and CPU temp with a fan on the Ninja.

I forgot to ask, what BIOS version (motherboard) do you have installed? And, what is the video card driver you are using for that 9600 GT?

tehcrazybob
Friend of SPCR
Posts: 356
Joined: Wed Jan 16, 2008 8:56 pm
Location: Council Bluffs, Iowa
Contact:

Post by tehcrazybob » Mon Apr 21, 2008 9:59 am

JVM wrote:I should have also mentioned that I am quite impressed you accomplished all this without putting a fan on the Ninja. Of course, you could most likely reduce your mainboard and CPU temp with a fan on the Ninja.
Thanks! I know I could stick an extra fan in, but it's been completely stable so I'm content. I'm also not too concerned with knowing the exact chipset temperature, but I suppose I'll take a look later today and let you know what it's at.

My BIOS is version F5 and my video drivers are the latest available directly from NVidia, 174.74.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Mon Apr 21, 2008 12:12 pm

tehcrazybob wrote:
JVM wrote:I should have also mentioned that I am quite impressed you accomplished all this without putting a fan on the Ninja. Of course, you could most likely reduce your mainboard and CPU temp with a fan on the Ninja.
Thanks! I know I could stick an extra fan in, but it's been completely stable so I'm content. I'm also not too concerned with knowing the exact chipset temperature, but I suppose I'll take a look later today and let you know what it's at.

My BIOS is version F5 and my video drivers are the latest available directly from NVidia, 174.74.
I am very interested to know what temp you get for the chassis/mainboard. Give that Hmonitor a try, it should work for you.

I assume you had to update your BIOS to F5?

tehcrazybob
Friend of SPCR
Posts: 356
Joined: Wed Jan 16, 2008 8:56 pm
Location: Council Bluffs, Iowa
Contact:

Post by tehcrazybob » Mon Apr 21, 2008 12:38 pm

The board came with the F5 BIOS already in place, actually.

I've also taken a look at HMonitor - it shows exactly the same temperatures and sensors as SpeedFan. In both programs, there are two that might help you.

The first sensor, which HMonitor reports as Mainboard and SpeedFan calls "Temp2," runs between 23c and 27c depending on load. I believe this is a thermal diode on the motherboard near the PCIe socket.

The second, which HMonitor identifies incorrectly as CPU1 and SpeedFan calls "Temp1", stays at 40c or 41c all the time. This might be the northbridge, but from touching the heatsink I'd guess it's actually hotter than that.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Mon Apr 21, 2008 1:01 pm

I seriously doubt that 23-27C for the mainboard. :)

You have a dual-core CPU and likely need Real Temp or Core Temp to get more accurate readings for your CPU; on the other hand, the E8400 is known to have a sensor problem in many of their chips and report inaccurate temps.

I don't know why Hmonitor would give bad readings for your mainboard, but over 40C at idle would have me a little concerned, if that is really your mainboard temp and your room is not warm. Would be interesting if you mounted a fan on the Ninja and checked those temps again, and using Real Temp or Core Temp to read your CPU temperature, although you may still get a false reading due to defective sensor in that E8400 chip.

tehcrazybob
Friend of SPCR
Posts: 356
Joined: Wed Jan 16, 2008 8:56 pm
Location: Council Bluffs, Iowa
Contact:

Post by tehcrazybob » Mon Apr 21, 2008 4:42 pm

JVM: I'm afraid I disagree with everything you just said.

First, the sensor being called 'mainboard' is literally the temperature of the motherboard. Not of any chip or processor, just the board itself. It's a thermal diode somewhere near the PCIe slot, and it measures the temperature of the free air inside the case. Considering that my room temperature is 23c and a quick feel inside suggests the case isn't much warmer than that, 23-27c is completely reasonable.

Second, nowhere has it been said that I have an E8400; I have an E6550. The E6000-series don't have broken sensors, although certain models read as 15c too low. Mine originally did but I applied a 15c offset in SpeedFan to match the readings from CoreTemp.

Third, HMonitor doesn't seem to be giving bad readings; it's giving me the same readings as SpeedFan and although HMonitor is labeling them incorrectly they are still valid measurements. The only one that's even slightly questionable is that 40-41c, and I'm pretty sure that's the northbridge temperature. That's not just the idle temperature, either; it's the total range of temperatures no matter what the load conditions. Considering that, 41c is comfortably safe. Chipsets and video cards both generally run warmer than CPUs, and even CPUs are perfectly comfortable at 41c.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Mon Apr 21, 2008 5:22 pm

Sorry, I thought you had the E8400 CPU processor.

As for the "motherboard" temperature, I use Hmonitor and generally get 35-36C with a room temperature of 21-22C. So, I am certainly surprised to see you get 23-27C using same software and with same case. If the Ninja is making that much of a difference, I would be very surprised. I hope you are not going by the temperature that displays after just a few minutes and have the computer on for at least 20 minutes.

Considering your CPU and GPU temps are identical to mine, I am doubly surprised that you get more than 10C less on what Hmonitor calls Mainboard.

That Ninja must be one heck of a cooler! :)

tehcrazybob
Friend of SPCR
Posts: 356
Joined: Wed Jan 16, 2008 8:56 pm
Location: Council Bluffs, Iowa
Contact:

Post by tehcrazybob » Mon Apr 21, 2008 7:01 pm

The Ninja is indeed a heck of a cooler, but there's more to it than that. The Ninja doesn't blow hot air everywhere, and the mainboard sensor is upwind of it. As a result, the heat from the CPU isn't having any effect on the reading at all. With a normal top-down cooler, the hot air off the cooler gets blown over the board. It's cooler than voltage regulators and the northbridge, so the airflow helps cool those parts. It's warmer than most of the other parts of the board, though, so it'll actually heat those up. Your mainboard temperature sensor might be experiencing this.

Aside from that, though, our 'mainboard' sensors may be in totally different places.

HammerSandwich
*Lifetime Patron*
Posts: 1288
Joined: Sat Oct 25, 2003 3:21 pm
Location: 15143, USA
Contact:

Post by HammerSandwich » Tue Apr 22, 2008 5:35 am

Folks, this conversation is far off-topic.

JVM
*Lifetime Patron*
Posts: 1564
Joined: Fri Nov 07, 2003 8:44 pm
Location: USA

Post by JVM » Tue Apr 22, 2008 5:50 am

HammerSandwich wrote:Folks, this conversation is far off-topic.
Very true, and I am sorry to have taken this thread off course. I accept full blame as I was the instigator. I shall desist from continuing any further discussion off the main topic.

Sorry for the inconvenience. :oops:

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Tue Apr 22, 2008 5:50 am

So bringing things back on topic, I think my Leadtek 9600GT with an aftermarket Accelero S1 and custom VRM heatsink is the best silent 9600GT so far:
viewtopic.php?p=408544#408544
Give it separate 2D/3D speeds and voltages and it would be better.

Post Reply