| HIS Radeon HD6950 IceQ-X Turbo-X Video Card |  |
| Reviews - Featured Reviews: Video Cards | |
| Written by Steven Iglesias-Hearst | |
| Thursday, 09 June 2011 | |
|
Page 3 of 18
HIS HD6950 IceQ X Turbo X DetailsIn this section we shall take an in-depth look at the HIS HD6950 IceQ X Turbo X video card and see what makes it tick. 
With the cooler assembly and other heatsinks fully removed we can get a better look at the board, The overall layout of all the components is a little different with the 6000 series, the standard layout has moved the power phase/VRM section to the left hand side of the GPU, into an area that is normally left somewhat unoccupied. All in all the PCB looks good with no real waste of space and the soldering quality is of a very high standard as you will see in the close-up shots further down the page.
The back of the PCB is utilised mainly for resistors and the soldering quality is excellent for such tiny components, man loses the war to the machine when it comes to detailed work like this. These days you don't generally see RAM on the reverse side of a 2GB video card design thanks to the smaller manufacturing process that allows a higher density in a smaller package. There are a few points of interest that we will look at in more detail in this section.
The HIS HD6950 IceQ X Turbo X uses 2GB of Hynix H5GQ2H24MFR-T2C GDDR5 Memory rated 1250MHz (5GHz effective) at 1.5V. On this model they are already running over spec (at H5GQ2H24MFR-R0C speeds).
For voltage control HIS have utilized a CHiL CHL82414-05 Dual output 4+1 phase PWM Controller. Below is a snippet from the product description. The CHL8212/13/14 are dual-loop digital multi-phase buck controllers and the CHL8203 is a single-loop digital multiphase buck controller designed for GPU voltage regulation. Dynamic voltage control is provided by registers which are programmed through I2C and then selected using a 3-bit parallel bus for fast access. The CHL8203/12/13/14 include CHiL Efficiency Shaping Technology to deliver exceptional efficiency at minimum cost across the entire load range. CHiL Dynamic Phase Control adds/drops active phases based upon load current and can be configured to enter 1-phase operation and diode emulation mode automatically or by command. SOURCE: chilsemi.com.
A uP6101BU8 is somewhat of an older chip seen on much earlier 4000 series AMD Radeon video cards, provides voltage control for the memory.
Two uP7701U8 (above and below) voltage control chips are present, one on the front and one on the reverse of the PCB. Specifications for these chips are not readily available but there is a good chance that they are an improvement on an older design. They provide voltage control for the GPU.
One can only speculate as to why there are two controllers but they may not be both performing the same task. When we look at the vital statistics of this video card in GPU-Z we see lots of extra info not normally seen on this level of card. For instance, instead of just a regular vCore reading we get readings for 12V line quality, VDDC usage in volts and VDDC current usage in amps.
The Pericom PI3HDMI4 chip controls the HDMI and DVI interfaces. Seen on a lot of Radeon video cards, this is likely to be the main reason why AMD video cards can support more than two monitors whereas NVIDIA cards can only support two max.
|
|
Comments
I'm glad to be different from the masses, I do a lot of research before I buy things and I like to get the best deals possible.
Your comparison charts spoke of something, to me at least. All the cards in the comparison would be pleasing to own. Not a real "dog" amongst them. I'm keeping in mind that gaming performance is a primary consideration. But even without games, these cards are all excellent to have in an HTPC or demanding video environment.
Oh and one more thing to watch when buying HIS graphics cards is the Fan wont last long if it's anything like the one on my HIS HD5770 (which it looks to be) not even 6 months old and it's already noisy as on start-up takes about 10~15mins of running before it gets to some semblance of quiet.. You all know the sound that grinding labored sound a fan starts to make when it's just about to fail
try it at 1920x1080 with lower AA and see how it goes
For the benchs I seen of one of those cards fully unlocked and with some good extra OC put in, its performance is way better than 570.
Unfortunately, these low-profile manufacturers are so for a reason, they don't have any feature to stand themselves ought of several sharks like PowerColor, Sapphire, XFX, ASUS and of course MSI.
In the end you only need that super high speed fan on TF3 if you're OCing it close or beyond a 1ghz core frequency.
But please note this are just suppositions I've made base on looking at both coolers and reading testimonies. For example, notice that HIS cooler, unlike TF3, hasn't got the back opened, allowing to suck in the air coming from front fans.
P.S.: For instance, because of that feature alone, I'm thinking on making a mod to my case so I can mount a Noctua P12 fan close to HD bays, to pressure air into the Graphics card cooler through the back of the cooler (the 6 pin connectors are above, so the intake from that position is very optimized)
And you completely ignored what I said about the noise per watt of heat spread. Without keeping this in mind you're just comparing apples with oranges...
Your guesses and suppositions might be plausible in the absence of any concrete data, but reviewer Steven Iglesias-Hearst did in fact point you to the concrete data that informed his assessment, data that he derived from (1) Full Reviews of the two cards that (2) He conducted personally, 17 days apart, using (3) A single Temperature Testing methodology (FurMark 1.9.0 Torture Test) and using (4) The exact same 'Lancool PC-K63'/'Intel P55'/'i5 760' System for each Review.
It?s rare that so many variables (Reviewer, Methodology, Case and System Cooling Config) are kept constant when comparing two cards, so his assessment of the relative merits of the two Cards should probably carry more weight than random testimonials. His results:
HIS 6950 IceQ X - Ambient Temp: 24 / Fan: 100% / FurMark Load Temp: 61 / Sound: ?Quite Bearable? Noise is a non-issue?
MSI 6950 Twin Frozr III -- Ambient Temp: 28 / Fan: 100% / FurMark Load Temp: 59 / Sound: ?Noise level is too much??
Granted, (5) Ambient temperatures are different (24 versus 28 degrees), but if (6) the Twin Frozr Fan is creating significantly more noise than the IceQ X, while (7) achieving approximately the same Load Temperature in FurMark (59 versus 61 degrees), then (8) your speculation -- ?You would need a lower speed [on the Twin Frozr 3] to match the [HIS IceQ X], and at that lower fan speed, it would make less noise than the top HIS heat spread? ? is not really consistent with the data.
Note also that in the test, the HIS Card was running at a higher default clock speed (880 MHz, versus 850 MHz for the MSI Card).
To be sure, we could ask Steven Iglesias-Hearst to (9) quantify (in decibels) ?quite bearable? and ?too much?, and also to (10) demonstrate the effect of a 4 degree difference on Ambient Temperature on FurMark Load Temps, but I think the data he has already provided is more than enough to support his assessment that the HIS Cooler is ?better? overall.
In the absence of such survey data, the overclocking results in Hearst?s review -- Street Fight IV 152.76 FPS @ 950/1350 (max OC) for the HIS IceQ X, versus Street Fighter IV 143.30 FPS for the MSI Twin Frozr III @ 900/1325 (max OC) -- give us the most helpful picture of the relative power/performance of the two cards.
Those OC numbers (900/1325) aren't the mirror of that MSI's true potential, since I've read dozens of high OCing numbers, being 975mhz on the core what most got, and a few lucky individuals got 1000mhz+ (cant remember memory speed). With such numbers I believe MSI's card would get similar results as HIS', and but with more durable and stable OC with its well-known military class components.
Here's win or win situation in which specific personal requirements make a difference (OC roof, silence, durability), but overall they seem to be in same level (opposite of my previous statements)
I end up loosing the right time for purchase with HD6xxx, so for HD7xxx I'll keep an open mind about this 2 coolers.
Offtopic: Does anyone think the thermal paste that comes with the card (with all brands) is as good as Arctic Silver 5 or a similar one? Never got the chance to compare then and now results with same ambient temperature, reason why I'm asking.
It's very hard to say without doing actual testing, but in my honest opinion you will not gain more that 2-3 degrees Celsius if you are lucky.
idle temps before AS5 were 36~37c after AS5 and curing time 27c
load temps before AS5 63c after AS5 52c (using F@H to load GPU's)
Because over time thermal paste gets dry and loses some of their heat transference capabilities, reason why if you already used the card for some time it's an unfair comparison.
I got the same results back then, after changing the 2 year old thermal paste of my HD4850, reason why I guess you did the same.
to flash my card is one that is the 840mhz version
thanks before hand
I need this card is at 880MHz core gpu bios
can use this bios
my original bios is the 840mhz version?