Radeon HD 4670: A perfect balance?

They make noise, too.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Vicotnik
*Lifetime Patron*
Posts: 1828
Joined: Thu Feb 13, 2003 6:53 am
Location: Sweden

Post by Vicotnik » Wed Sep 17, 2008 3:39 pm

Fungi wrote:As posted earlier, the HD 4670 holes are 43mm apart.
Sorry I must have missed that post. The holes on the S1 rev 1 are 54mm apart I think.

Fungi
Posts: 95
Joined: Fri Dec 01, 2006 1:27 pm

Post by Fungi » Wed Sep 17, 2008 9:27 pm

The results are in and I'm impressed. After I put in the order for the S1 I realized that the S2 would have been a better match for the 4670, being smaller and all. Based on results, I'd say the S2 is more than adequate. :lol:

First a word of warning, you have to slide the mounting bracket all the way back or the heatpipes will be hitting or really close to mobo components. But once slided it should be fine unless you have anything higher than the PCI-e bracket around it. This is just the reference design, of course, and other boards may move the chip around.

Anyway, ambient temps are the same or have risen because it's getting kind of warm in here but the GPU is 35C idle 46C load now. In comparison, the Sapphier cooler with stock fan settings (loud) got 40C idle 65C load in my system.

Picture of the system, note the Nexus blowing directly on it.
http://img233.imageshack.us/img233/5934/systemks1.jpg
Last edited by Fungi on Thu Sep 18, 2008 8:43 am, edited 1 time in total.

Max Dread
Posts: 82
Joined: Thu Aug 21, 2008 6:36 am
Location: Norfolk, UK

Post by Max Dread » Wed Sep 17, 2008 11:48 pm

You're a star Fungi - great help and many thanks.

Just to clarify, are you saying the processor on the card is set quite low on the PCB? And that as a result, the accelero sits lower and therefore closer to the motherboard? I have an Intel DP35DP and there are some caps and a small heatsink the sit around 5-10mm higher than the PCIe slot. Do you thinl this will be a problem?

Cheers

Max

shathal
Posts: 1083
Joined: Wed Apr 14, 2004 11:36 am
Location: Reading, UK

Post by shathal » Thu Sep 18, 2008 8:49 am

Fungi wrote:The results are in and I'm impressed. After I put in the order for the S1 I realized that the S2 would have been a better match for the 4670, being smaller and all. Based on results, I'd say the S2 is more than adequate. :lol:

First a word of warning, you have to slide the mounting bracket all the way back or the heatpipes will be hitting or really close to mobo components. But once slided it should be fine unless you have anything higher than the PCI-e bracket around it. This is just the reference design, of course, and other boards may move the chip around.

Anyway, ambient temps are the same or have risen because it's getting kind of warm in here but the GPU is 35C idle 46C load now. In comparison, the Sapphier cooler with stock fan settings (loud) got 40C idle 65C load in my system.

Picture of the system, note the Nexus blowing directly on it.
http://img233.imageshack.us/img233/5934/systemks1.jpg
Lol - nice picture - talk about a gfx card being dwarfed by its heatsink :).

Great work Fungi!

Fungi
Posts: 95
Joined: Fri Dec 01, 2006 1:27 pm

Post by Fungi » Thu Sep 18, 2008 10:27 am

Max Dread wrote:You're a star Fungi - great help and many thanks.

Just to clarify, are you saying the processor on the card is set quite low on the PCB? And that as a result, the accelero sits lower and therefore closer to the motherboard? I have an Intel DP35DP and there are some caps and a small heatsink the sit around 5-10mm higher than the PCIe slot. Do you thinl this will be a problem?

Cheers

Max
If it's where the heatpipes will go, it might be a problem. I'll take some more pictures and measurements later.

daaron
Posts: 6
Joined: Mon Aug 21, 2006 3:22 am
Location: California

Post by daaron » Fri Sep 19, 2008 7:18 pm

I recently bought the Sapphire 4760, as well as the AcceleroS1 rev2.

I just wanted to test out the stock cooler, and stuck it into my system. The fan might look small, but the noise it creates isn't. It's sorta like a mini vacuum cleaning going off inside of my PC.

Before I install the Accelero, I wanted to get a sort of "baseline" on the temps off this card, so I started up Speedfan. To my surprise, the temp sensors did not show up. I only had my mobo, cpu, and hard drives show up.

Since Fungi just recently did an install, I was wondering if you ran into this problem? Or anyone really, if they can help me figure out what's going on.

On a side note, this is an upgrade from using the integrated Radeon 3200. I used to average about 35 FPS in CS:S, and now push approximately 70ish. As a budget card, this seems like a pretty decent increase.

Vicotnik
*Lifetime Patron*
Posts: 1828
Joined: Thu Feb 13, 2003 6:53 am
Location: Sweden

Post by Vicotnik » Fri Sep 19, 2008 7:46 pm

You can see temps with CCC, ATITool and RivaTuner I think. I usually use CCC which just shows you the temp that instant, but with the other two you can play a game and then look at a graph how the temp changed over time.

I also has a Sapphire 4670 and a S1 rev2 on the way. :)

daaron
Posts: 6
Joined: Mon Aug 21, 2006 3:22 am
Location: California

Post by daaron » Fri Sep 19, 2008 8:06 pm

Hmm, I just installed the drivers/CCC off the CD that came with it, as 4760 drivers aren't up on the ATI site yet. To be honest, I did look through the CCC, advanced view or basic view. Maybe I'm just suffering from too much stupidity, but I couldn't figure out any way to see my temps with the CCC. However, I did go to guru3d, and downloaded GPUZ, which DOES show me current temps.

Freakily enough, using GPUZ, when I look at temperatures, I see about 45C on average for my GPU temperature. However, my fan speed will read 24%, and say 1RPM. I can tell that it's definitely not the case. As well, my GPU temp for the shadercore reads at 349C :O
If I weren't so sure that that number was complete bullshit I'd be panicking.

Unfortunately, I'm not sure if GPUZ is capable of graphing temperatures for me. Anyways, next step is to then see what I can do with an Accelero.

Thanks for helping out.

Fungi
Posts: 95
Joined: Fri Dec 01, 2006 1:27 pm

Post by Fungi » Fri Sep 19, 2008 8:14 pm

I'm using RivaTuner (latest) to monitor temps. Latest version added support for the 4670.

By the way, is anyone's CCC.exe crashing? (I just disabled it since I don't need it.)

daaron
Posts: 6
Joined: Mon Aug 21, 2006 3:22 am
Location: California

Post by daaron » Fri Sep 19, 2008 8:27 pm

Mine's seems to run ok? No crashing at all, but then again I haven't really needed to use it for anything.

As a question to fungi: I only see 4 places to stick my ram sinks, by your pictures, it seems you've stuck sinks on both sides. Did you end up using any of the voltage sinks? As I understand it, the 4670 uses a relatively low amount of power, so there is probably no need, but I was wondering if this would impact the videocard at all?

Thanks.

Fungi
Posts: 95
Joined: Fri Dec 01, 2006 1:27 pm

Post by Fungi » Fri Sep 19, 2008 11:19 pm

There are 4 ram chips on each side, in the exact same spots. It's GDDR3 though, so you probably don't *need* the sinks. And I didn't use any of the voltage sinks, didn't use them. Didn't really see anything that seemed like it.

Tobias
Posts: 530
Joined: Sun Aug 24, 2003 9:52 am

Post by Tobias » Sun Sep 21, 2008 5:02 am

Has anyone found any info on the power draw of this card yet? Trying to compare the numbers in the reviews only makes one crazy as they in some cases draw 15W more than some other card and in some cases 12W less than the same chipset.

Kriz
*Lifetime Patron*
Posts: 51
Joined: Tue Sep 02, 2008 10:16 am
Location: Australia

Post by Kriz » Sun Sep 21, 2008 10:43 am

Regarding power draw, I recently added a HIS HD4670 IceQ 4 Turbo 512MB to the following configuration:

viewtopic.php?p=429214#429214

I tested this system with the E8400 at stock speed, undervolted to 1.1v in BIOS. PSU is a Seasonic SS-300SFD 300W.

Using onboard GPU:
41W Idle

Using HIS HD4670:
54W Idle / 108W Peak while playing DiRT

I am very happy with this card.

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Post by ~El~Jefe~ » Sun Sep 21, 2008 11:33 am

unless you have a 1920x1200 monitor, that 4670 looks awesome.

i only trust hexus numbers and power draw btw

Vicotnik
*Lifetime Patron*
Posts: 1828
Joined: Thu Feb 13, 2003 6:53 am
Location: Sweden

Post by Vicotnik » Sun Sep 21, 2008 2:22 pm

~El~Jefe~ wrote:i only trust hexus numbers and power draw btw
Why Hexus? I think they report a very high idle power consumption for the 3850 for instance.

Light Yagami
Posts: 8
Joined: Sat Sep 06, 2008 3:35 pm
Location: Sweden

Post by Light Yagami » Sun Sep 21, 2008 4:05 pm

~El~Jefe~ wrote:unless you have a 1920x1200 monitor, that 4670 looks awesome.
Why do you say this? I'm considering an HD4670 and intended to get a 1920x1200 monitor, as well. Is there something I should have in mind before I get this card (if I do)?

nici
Posts: 3011
Joined: Thu Dec 16, 2004 8:49 am
Location: Suomi Finland Perkele

Post by nici » Sun Sep 21, 2008 4:38 pm

Mr. Jefe is saying that it might not be adequate if you do a lot of gaming running the aforementioned resolution. The 4670 will have no trouble dispaying 1900x1200.
Or maybe he's saying something completely different, in which case i'll just have another glass of Ardbeg in preparation of my crucifixion. On second thought, i'll just have it anyway. Just in case :wink:

Vicotnik
*Lifetime Patron*
Posts: 1828
Joined: Thu Feb 13, 2003 6:53 am
Location: Sweden

Post by Vicotnik » Sun Sep 21, 2008 5:07 pm

nici wrote:Mr. Jefe is saying that it might not be adequate if you do a lot of gaming running the aforementioned resolution. The 4670 will have no trouble displaying 1900x1200.
That is the way I interpreted it also.

The 4670 can even display 2560x1600 since it has Dual-Link DVI. I agree with ~El~Jefe~ that the card is a little bit to slow for serious gaming at 1900x1200. I am aware of that and I'm ok with it since I seldom play newer games and if I do I can use 1680x1050 and let my TFT scale up the picture, or play at native 1900x1200 and lower the IQ settings a bit more. Whichever looks best. :)

So get that 4670 Light Yagami. 8)

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Post by ~El~Jefe~ » Mon Sep 22, 2008 6:07 am

nici wrote:Mr. Jefe is saying that it might not be adequate if you do a lot of gaming running the aforementioned resolution. The 4670 will have no trouble dispaying 1900x1200.
Or maybe he's saying something completely different, in which case i'll just have another glass of Ardbeg in preparation of my crucifixion. On second thought, i'll just have it anyway. Just in case :wink:
yeah. hi Nici :D

It could not handle a 2007+ game at 1920x1200. lower resolutions it looks great. Oh and hexus? they are smart cookies over there. I would also look at relative power usage not the exact numbers. The only accurate power usage reviewing site is this one. We just do not do reviews of such things it seems.

dhanson865
Posts: 2198
Joined: Thu Feb 10, 2005 11:20 am
Location: TN, USA

Post by dhanson865 » Mon Sep 22, 2008 6:47 am

On power draw I found this today.

The Truth About Grapics Power Requirements V2

I'm not going to post the whole thing here but I will quote a small portion
--------------------------------------------------
GFX Card Comparison Table
--------------------------------------------------

This table shows GFX cards power consumption, not what the recommended PSU requirements are, but rather the actual draw of the card. The graphs show max 3D power draw as well as idle and max 2D power draw where possible.

All readings have been found on the web, you can find where I got the details from by looking in the "source links" legend below the graphs. I'v provided a simple legend to help you find where it came from. (eg look for an *, ², ^ etc after each reading)

The measurements displayed after each bar on the graph are displayed as min power draw - max 2d power draw - max 3d power draw . If there's only two readings it's likely just min-max 3D, and if there's just the one reading it'll be the max 3d reading.

Last thing of note. I've tried to normalize all the readings found at the various websites against cards that have been properly tested for power draw (eg what xbitlabs mostly does), not just a generic total system score. This should hopefully provide more accurate results for cards with readings that are otherwise of not much use.


For nice and pretty picture versions check these out (kept up to date):
Charts sorted by peak 3D - http://mark.zoomcities.com/images/gfx/G ... rtby3d.png
Charts sorted by idle - http://mark.zoomcities.com/images/gfx/G ... byidle.png


Graph Legend
-------------------
# = 8 Watts
¦) = 8 Watts (min draw mark)
¦] = 8 Watts (max 2D draw mark)

ATI Single Card
----------------------------

HD4870 X2____________##########¦)#####################}| 88W-262W !
HD4870 ______________########¦)¦]######| 65W-80W-130W ²
HD4850 X2____________############################}| 230W %
HD4850 OC____________#######¦)########}' 64W-133W § (800/1250)
HD4850 ______________#####¦)#¦]#####}| 41W-61W-110W ²
HD4670 _________#¦)######}| 10W-70W %
HD4650 ______________#¦)######## 10W-80W %
HD4470 ______________#¦)####| 10W-50W %
HD4450 ______________#¦)#}| 10W-30W %

HD3870 X2____________#######¦)¦]#############|' 58W-66W-171W ²
HD3870 ______________##¦)¦]######' 19W-25W-81W ²
HD3850 X2____________######¦)###########' 54W-145W §
HD3850 ______________#¦)¦]####}|' 14W-20W-63W ²
HD3690 ______________##¦)###}' 21W-51W §
HD3650 ______________#¦)###| 12W-42W ² (512MB)
HD3650 ______________##¦)###}' 19W-53W § (256MB)
HD3470 ______________#### 32W
HD3450 ______________##¦)#} 18W-36W §

HD2900 XTX___________#################################}| 270W % (OEM 12.4")
HD2900 XTX___________############################## 240W % (Retail 9.5")
HD2900 XT____________#######¦)##########}| 65W-150W ! (1GB GDDR4)
HD2900 XT____________########¦)¦]##########' 72W-76W-161W ² (512MB GDDR3)
HD2900 Pro___________########¦)¦]######}' 71W-76W-133W ²
HD2900 XL____________######################} 180W %
HD2600 XT____________##¦)¦]##' 21W-28W-49W ²
HD2600 Pro___________#¦)¦]}|' 16W-22W-31W ²
HD2400 XT____________¦)¦]|' 7W-15W-19W ²
HD2400 Pro___________¦)}|' 9W-15W &
HD2400 LE ___________###' <25W +

X1950 XTX____________####¦)#¦]########}' 33W-54W-125W ²
X1900 XTX____________###¦)##¦]########' 29W-50W-121W ²
X1950 XT_____________############ 96W
X1900 XT_____________###¦)#¦]#######}' 28W-46W-109W ²
X1900 XT_____________###¦)########} 26W-100W ^ (R580+ 256MB model)
X1950 Pro____________##¦)#¦]###| 24W-36W-66W ²
X1950 GT ____________###¦)###' 30W-57W &
X1900 GT ____________###¦)#¦]###|' 25W-45W-75W ²
X1800 XT OC _________###¦)#¦]######## 26W-48W-110W ² (XT OC version, 700/1600)
X1800 XT_____________###¦)##¦]#####}|' 30W-52W-103W ²
X1800 XL_____________###¦)#¦]#' 27W-42W-57W ²
X1800 GTO___________###¦)¦]# 25W-35W-48W ²
X1650 XT_____________##¦)##¦]}|' 23W-42W-55W ²
X1650 Pro____________#¦)###} 23W-44W &
X1600 XT_____________##¦)#¦]| 24W-38W-42W ²
X1600 Pro____________##¦)¦]#' 24W-30W-41W ²
X1550 _______________###|' 27W
X1300 XT_____________##}| 22W
X1300 Pro____________##¦)}|' 18W-26W-31W ²
X1050 _______________### 24W

X850 XT PE___________########|' 67W
X850 XT______________###¦)#¦]##}' 27W-45W-69W ²
X800 XT PE___________##¦)####}|' 18W-63W ²
X800 XT ______________######}| 54W
X800 Pro _____________#¦)#### 15W-48W ²
X800 XL ______________##¦)#¦]#' 18W-36W-49W ²
X800 GTO ____________##¦)¦]##' 19W-31W-49W ²
X800 GT______________##¦)¦]# 21W-31W-40W ²
X800_________________###}| 30W
X700 XT______________#¦)##}| 12W-38W !
X700 PRO ____________#¦)#¦]' 16W-30W-33W ²
X700_________________#¦)###} 15W-44W ^
X600 XT______________#¦)#' 11W-25W !
X600 Pro_____________#¦)##} 12W-36W ^
X300_________________¦)###} 8W-36W !
X300 SE______________¦)##}| 7W-30W ^

9800 XT______________####¦)##} 38W-60W ²
9800 Pro_____________###¦)#}|' 31W-47W ²
9700/9800____________####| 34W
9600 XT______________#¦)}| 9W-22W ²
9600 Pro_____________#¦)| 9W-18W ²
9500/9600____________#} 12W
9000-9200____________###} 28W

7x00-8500____________##}|' 23W
<=32MB AGP/PCI _____#}|' <15w


nVidia Single Cards
---------------------------------
7950 GX2 ____________######¦)######}| 50W ^-110W ²
7900 GX2 ____________#############}' 109W
7900 GTX ____________###¦)##¦]###} 31W-52W-84W ²
7900 GTO ____________###¦)#######} 27W-92W &
7950 GT _____________###¦)¦]##}' 26W-37W-61W ²
7900 GT _____________##¦)¦]## 23W-32W-48W ²
7900 GS _____________##¦)¦]#}' 19W-28W-45W ²
7800 GTX 512_________###¦)##¦]####}' 29W-53W-95W ²
7800 GTX ____________###¦)##¦]###' 29W-52W-81W ²
7800 GT _____________##¦)#¦]##' 20W-39W-57W ²
7800 GS _____________#####} 44W
7600 GT _____________#¦)¦]#} 15W-23W-36W ²
7600 GS _____________#¦)¦]|' 14W-20W-27W ² (underclocked GT)
7300 GT _____________#¦)#}' 13W-29W &
7300 GS _____________#¦) 9W-10W-16W ²
7300 LE______________#| 10W

6800 Ultra____________###¦)##### 29W-72W ²
6800 GT _____________##¦)####' 23W-55W ²
6800 GS_____________##¦)##¦]}|' 20W-41W-55W ²
6800 ________________#¦)#¦]}|' 15W-31W-39W ²
6600 GT _____________##¦)#¦]# 19W-39W-48W ²
6600 ________________#¦)¦]|' 12W-24W-27W ² (GDDR2 model)
6200 ________________#¦)#| 11W-26W !

FX 5950 Ultra_________##¦)######' 20W-73W ²
FX 5900 Ultra_________###¦)###|' 27W-59W ²
FX 5900 XT___________##¦)#|' 24W-35W !
FX 5800 _____________#####} 44W
FX 5700 Ultra_________##¦)##}| 24W-46W ²
FX 5700 _____________#¦)# 10W-24W ²
FX 5600 _____________####}' 37W
FX 5200 _____________###|' 27W

GF4 Ti ______________##### 40W
GF4 MX _____________##}| 22W
GF3 Ti ______________####|' 35W
<=32MB AGP/PCI_____#}|' <15w

Quadro range

FX 5600______________#################| 138W
FX 5500______________#########} 76W
FX 4500______________########### 88W
FX 4400______________#######' 57W
FX 4000 PCI-E ________########}|' 71W
FX 4000 AGP _________######### 72W
FX 3700______________#¦)#######|' 14W-75W ²
FX 3450______________########}| 70W
FX 3400______________######}| 54W
FX 1700______________¦)######}' 7W-61W ²
FX 1400______________##¦)######|' 17W-75W §
FX 1300______________######|' 51W
FX 570_______________¦)######' 7W-57W ²
FX 540_______________###}' 29W
FX 330_______________###}| 30W

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Post by ~El~Jefe~ » Mon Sep 22, 2008 7:19 am

that's whacky

so, my 3870 doesnt even use 90 watts. hm.

QuietOC
Posts: 1407
Joined: Tue Dec 13, 2005 1:08 pm
Location: Michigan
Contact:

Post by QuietOC » Tue Sep 23, 2008 12:30 pm

I had every attention of ordering a Radeon HD 4670 until I saw a $50 Geforce 9600 GSO with a nice dual-slot cooler. I do have to wait/trust for the rebate from EVGA though.

I am sucker for these down market high-end parts--having purchased a Radeon X800 GTO when the Geforce 7600 GT was all the rage. The Geforce 9600 GSO is basically a 3/4ths GeForce 9800 GTX. This one even has the GTX cooler.
Last edited by QuietOC on Tue Sep 23, 2008 12:35 pm, edited 1 time in total.

Auroa
Posts: 27
Joined: Sun Jan 23, 2005 8:41 pm

Post by Auroa » Tue Sep 23, 2008 12:34 pm

FWIW: http://www.bit-tech.net/hardware/2008/0 ... 0-512mb/11

Idle power consumption of the 4670 is strangely higher than the 3800s.

QuietOC
Posts: 1407
Joined: Tue Dec 13, 2005 1:08 pm
Location: Michigan
Contact:

Post by QuietOC » Tue Sep 23, 2008 12:55 pm

Auroa wrote:Idle power consumption of the 4670 is strangely higher than the 3800s.
Cards based on these GPUs will vary a lot--especially idle power. Maybe the BIOS isn't as aggressive in lowering voltages? They could even be overvolting cheaper RAM (the whole DDR3 is as good as GDDR3 scares me.) The GPU chips themselves are going to have sample variance. From a simple physics perspective a HD 4670 card should be able to use less power than a HD 3850 card--half as many bus lines to power at least. The power usage of RV730 should be pretty similar to the RV670 unless texture units or ROPs are big factors.

That Australian site with the numbers compiled may have a typo on the HD 3850 listing. They have two sets of numbers on the charts. The 256 MB 3850 being 14 W idle 63 W load, and the 512 MB HD 3850 being 20 W idle and 68 W load.

Vicotnik
*Lifetime Patron*
Posts: 1828
Joined: Thu Feb 13, 2003 6:53 am
Location: Sweden

Post by Vicotnik » Tue Sep 23, 2008 1:48 pm

Auroa wrote:FWIW: http://www.bit-tech.net/hardware/2008/0 ... 0-512mb/11

Idle power consumption of the 4670 is strangely higher than the 3800s.
Not mine. My system with a Sapphire 4670 draws just a little bit less than the same system with a HIS 3870. And that's with a modded BIOS on the 3870 that lowers the idle consumption by a few watts.

lechuck
Posts: 86
Joined: Mon Dec 04, 2006 11:57 pm
Location: EU

Post by lechuck » Wed Sep 24, 2008 2:27 am

Finally there may be hope for a new card with low power consumption in idle/2D mode!

I'd ask users who own 4670 cards to write how much the card underclocks in 2D mode.
If they can measure power consumption, would these figures be welcome.

I have 2600XT and I lowered idle power draw for about 10W with BIOS mod, so my card needs about 17W in idle.
Would be interesting to know how much 4670 did improve in this area.

panita
Posts: 7
Joined: Wed Sep 24, 2008 6:07 am
Location: UK

HIS HD4670 IceQ

Post by panita » Wed Sep 24, 2008 6:20 am

Kriz,

It looks like your 13W idle draw is a little higher that the charts in the reviews which seem to average out at about 10W. That maybe because of the small overclock on the HIS IceQ Turbo card.

I fact I'm looking for that card over in the UK at the moment but can't see any availability yet. Could you tell me how well the cooler works regards fan speeds and noise? Does it idle slowly and then speed up while playing games? If so how loud? Thanks....

Kriz
*Lifetime Patron*
Posts: 51
Joined: Tue Sep 02, 2008 10:16 am
Location: Australia

Post by Kriz » Wed Sep 24, 2008 8:05 pm

I'll have to do some testing with CPUBurn or Prime95 with and without the HD4670 to see if the GPU idle power is any different, like what was done in the recent Asus GTX260 review. I have a feeling this particular board uses more power when a PCIE video card is being used.

Currently the loudest part of this system is the unmodded Seasonic SS-300SFD, so I'll be putting the PicoPSU back in later tonight to see how the card sounds at idle at least, I'm not sure my 80w brick will like running any games just yet.

Hopefully RBE 1.15 will also allow me to list the clocks and voltage settings for this card.

digaderfox
Posts: 6
Joined: Wed Sep 24, 2008 1:14 pm
Location: France

Post by digaderfox » Wed Sep 24, 2008 9:34 pm

Hi dudes,

I need to ask you for some help because I run into an issue that totally puzzles me.

I bought two days ago two Connect3D 4670 512 (for a multiscreen setup).

I was disapointed by the noise of the fan as soon as I kicked it in. So I decided to run to the store and get two Accelero S1 rev2.

As for the mounting job, nothing special. Oh yes : there is a coil and a condenser that slightly touch the cooler in the right upper corner. Nothing bad : some tape to ensure isolation does a great job.

This is now thing goes wrong.
The idle temperature is at 34°.
While burning with Atitool, each time the temperature would reach 45° or so and Windows will reboot (Vista64). And each try is the same. It would never goes higher than 50°.

I tried with the two cards and results are the same.

So I rolled back to the little turbine. It is now running for half an hour without trouble.

I really can't understand what I'm missing.

It's cooler with Accelero but will hang with no reason ?

Could there be some kind of protection in the bios that makes it impossible for the card to run without a fan ? (I'll test the mixed solution)

Any advice would be appreciated.

Kriz
*Lifetime Patron*
Posts: 51
Joined: Tue Sep 02, 2008 10:16 am
Location: Australia

Post by Kriz » Fri Sep 26, 2008 1:32 am

I haven't had a chance to get the PicoPSU setup for testing, as I haven't been happy with the peak power use that this system is currently reading. Hopefully I should have more time after this weekend to plug in the NeoHE 430 to do some more subjective noise testing, but so far I can say that I haven't noticed any obvious noise difference between the card at idle or after 30 mins of playing DiRT.

According to RBE 1.15...

HD4670 idle clocks: 165MHz GPU / 250MHz RAM / 0.900v
HD4670 full speed: 780MHz GPU / 1000MHz RAM / 1.250v

Now for some more power testing with the Seasonic SS-300SFD 300W PSU...

Using onboard GPU: 41W Idle / 69W Load (Prime95 Small FFTs)
Using HIS HD4670: 54W Idle / 81W Load (Prime95 Small FFTs)

So, it's still around 12-13W extra for my particular configuration.

More to come next week.

Post Reply