The HD 4890 is ATI’s newest GPU, on step beyond the HD 4870. It is the fastest single GPU graphics card produced by ATI. We’re most interested in seeing if power consumption and noise level has improved over the 4870.
Apr. 13, 2009 by Lawrence Lee
HIS Radeon HD 4890 Turbo
PCI-E Graphics Card
(Hightech Information System Limited)
ATI’s recently dropped on their fastest single GPU graphics card, the HD
4870 to $149 and $199 for the 512MB and 1GB versions respectively. Their
manufacturer partners resisted the change, however, party due to decreased demand,
but also because the 4870 was already at a good price/position against nVidia’s
rival part, the GTX 260. The release of ATI’s newest GPU, the HD 4890, may push
them in line, lest the close pricing of the 4870 and the 4890 causes them to cannibalize
sales from one another.
Though early reports suggested the HD 4890 would simply be an overclocked HD
4870, this turned out not to be completely true. The 4890’s RV790 core is an
updated/tweaked version of the 4870’s RV770. The main difference is a set of
decoupling capacitors around the outside of the chip which results in a 10%
increase in physical surface area. This gives it the ability to handle more
voltage allowing for higher clock speeds. Timing and power distribution has
also been tweaked. Other changes include an increase of 100 MHz core clock speed,
75 MHz GDDR5 frequency, and a 30W higher power envelope. Our HD 4890 sample
is a "Turbo" edition made by HIS. It has a core clock speed of 900
MHz, 50 MHz higher than reference — several manufacturers have put out
similar factory overclocked 4890s.
Strangely, our sample came with no adapters or cables, but a copy of S.T.A.L.K.E.R.:
Clear Sky. According to most e-tailers, the retail version ships with adapters
for VGA and HDMI connectivity, a 6-pin power adapter, a CrossFire bridge and
no free games.
HIS Radeon HD 4890 Turbo Edition: Specifications
product web page)
The specifications list the card’s memory as being clocked at 1000 MHz (4000
MHz effective). According to the GPU-Z, our sample’s memory was running at 975
PHYSICAL DETAILS & INSTALLATION
The HIS Radeon HD 4890 has a reference design cooler, which incidentally is
pretty much a copy of the 4870 cooler.
Our test procedure is an in-system test, designed to:
1. Determine whether the card’s cooler is adequate for use in a low-noise system.
By adequately cooled, we mean cooled well enough that no misbehavior
related to thermal overload is exhibited. Thermal misbehavior in a graphics
card can show up in a variety of ways, including:
Any of these misbehaviors are annoying at best and dangerous at worst —
dangerous to the health and life span of the graphics card, and sometimes to
the system OS.
2. Estimate the card’s power consumption. This is a good indicator of how efficient
the card is and will have an effect on how hot the stock cooler becomes due
to power lost in the form of heat. The lower the better.
3. Determine the card’s ability to play back high definition video, to see
if whether it is a suitable choice for a home theater PC.
Measurement and Analysis Tools
Our first test involves recording the system power consumption using a Seasonic
Power Angel as well as CPU and GPU temperatures using SpeedFan and GPU-Z during
different states: Idle, under load with CPUBurn running to stress the processor,
and with CPUBurn and ATITool’s artifact scanner (or FurMark — whichever
produces higher power consumption) running to stress both the CPU and GPU simultaneously.
This last state mimics the stress on the CPU and GPU produced by a modern video
game. The software is left running until the GPU temperature remains stable
for at least 10 minutes. If artifacts are detected by ATITool or any other instability
is noted, the heatsink is deemed inadequate to cool the video card in our test
If the heatsink has a fan, the load state tests are repeated at various fan
speeds while the system case fan is left at its lowest setting of 7V. If the
card utilizes a passive cooler, the system fan is varied instead to study the
effect of system airflow on the heatsink’s performance. System noise measurements
are made at each fan speed.
Power consumption is also measured during playback of a variety of video clips
with PowerDVD to test the efficiency of the card’s H.264/VC-1 hardware acceleration.
Video Test Suite
Rush Hour 3 Trailer 1 is a H.264 encoded clip inside an Apple
Coral Reef Adventure Trailer is encoded in VC-1 using the WMV3
codec commonly recognized by the "WMV-HD" moniker.
VC-1: Drag Race is a recording of a scene from
network television re-encoded with TMPGEnc using the WVC1 codec, a more
demanding VC-1 codec.
Estimating DC Power
The following power efficiency figures were obtained for the
used in our test system:
Seasonic S12-500 / 600 TEST RESULTS
DC Output (W)
AC Input (W)
This data is enough to give us a very good estimate of DC demand in our
test system. We extrapolate the DC power output from the measured AC power
input based on this data. We won’t go through the math; it’s easy enough
to figure out for yourself if you really want to.
BASELINE, with Integrated Graphics: First, here are the results of
our baseline results of the system with just its integrated graphics, without
a discrete video card. We’ll also need the power consumption reading during
CPUBurn to estimate the actual power draw of discrete card later.
VGA Test Bed: Baseline Results
(no discrete graphics card installed)
Ambient temperature: 21°C
Ambient noise level: 11 dBA
System noise level: 12 dBA
ATI Radeon HD 4870 1GB:
VGA Test Bed: HIS Radeon HD 4890 Turbo Edition
CPUBurn + ATITool
CPUBurn + FurMark
Ambient temperature: 21°C.
Temperatures recorded via GPU-Z.
During testing, temperatures stayed well within reasonable levels — at full
load, the GPU core temperature measured only 75°C. The relatively low temperatures
were the result of the card’s cooling fan being more aggressive than we are
used to seeing. It sped up early and often once a load was applied to the GPU.
Idle, the fan spun at approximately 1400 RPM and measured 15 dBA. It was audible,
but soft sounding, like an optical drive spinning at a low speed. When the fan
speed stabilized on load, it was much louder. It measured 26 dBA and had a lot
more hiss due to higher air turbulence (though much of it was blocked
out by our case side panel), as well as a nasty growl.
VGA Test Bed: HD 4890 vs. HD 4870
HIS Radeon 4890 Turbo 1GB
ATI Radeon 4870 1GB
CPUBurn + ATITool
CPUBurn + FurMark
Ambient temperature: 21°C.
Temperatures recorded via GPU-Z.
Compared to our reference HD 4870 sample, it is clear the HD 4890’s fan control
is much more active, resulting in higher noise levels, but lower temperatures.
The reference 4870 could conceivably find a place in a quiet PC since it is barely audible
when idle. The same cannot be said of the reference 4890.
The power consumption of an add-on video card can be estimated by comparing
the total system power draw with and without the card installed. Our results
were derived thus:
1. Power consumption of the graphics card at idle – When CPUBurn is
run on a system, the video card is not stressed at all, and stays in idle mode.
This is true whether the video card is integrated or an add-on PCIe 16X device.
Hence, when the power consumption of the base system under CPUBurn is subtracted
from the power consumption of the same test with the graphics card installed,
we obtain the increase in idle power of the add-on card over the
integrated graphics chip (Intel GMA950). (The actual idle power
of the add-on card cannot be derived, because the integrated graphics does draw
some power — we’d guess no more than a watt or two.)
2. Power consumption of the graphics card under load – The power draw
of the system is measured with the add-on video card, with CPUBurn and FurMark
running simultaneously. Then the power of the baseline system (with integrated
graphics) running just CPUBurn is subtracted. The difference is the load power
of the add-on card. (If you want to nitpick, the 1~2W power of the integrated
graphics at idle should be added to this number.) Any load on the CPU from FurMark
should not skew the results, since the CPU was running at full load in both
Both results are scaled by the efficiency of the power supply (tested
here) to obtain a final estimate of the DC power consumption.
Power Consumption Comparison (DC)
Est. Power (Idle)
Est. Power (ATITool)
Est. Power (FurMark)
ATI HD 4830*
Diamond HD 4850
ATI HD 4870 1GB
HIS HD 4890 Turbo 1GB
* sample with unknown number of stream processors
Our 4890 sample had slightly higher power consumption than its predecessor,
the HD 4870, despite any power regulation improvements ATI may have implemented
in the new core, though it should be noted the Turbo edition card is factory
overclocked by 50 MHz. The power consumption of ATI’s faster cards hasn’t improved, not since the upgrade the 55 nm manufacturing process
(HD 3000 series).
Video Playback Power Consumption (AC)
Sparkle GTS 250 1GB
Diamond HD 4850 512MB
Asus ENGTX260 896MB
ATI HD 4870 1GB
HIS HD 4890 Turbo 1GB
Power consumption during video playback was similar as well. No surprises here.
Some of the card’s inner workings can be revealed by taking a look at the BIOS.
We used GPU-Z to extract the board’s BIOS and Radeon
BIOS Editor to examine its contents.
Displayed in green, yellow and red are the card’s settings for idle, UVD, and
3D. According to the BIOS, core voltage is 1.053V when idle and increases to
1.313V on load. We confirmed the idle and 3D frequency changes with GPU-Z, but
during video playback the card used the 3D clock speeds.
The fan control behavior follows a simple linear progression starting at a
GPU temperature of 25°C and finishing at 101°C. This is a very aggressive
setting — our reference 4870 does not increase its fan speed until the GPU
hits 58°C. There doesn’t seem to be a good reason for this change.
MP3 SOUND RECORDINGS
These recordings were made with a high
resolution, lab quality, digital recording system inside SPCR’s
own 11 dBA ambient anechoic chamber, then converted to LAME 128kbps
encoded MP3s. We’ve listened long and hard to ensure there is no audible degradation
from the original WAV files to these MP3s. They represent a quick snapshot of
what we heard during the review.
These recordings are intended to give you an idea of how the product sounds
in actual use — one meter is a reasonable typical distance between a computer
or computer component and your ear. The recording contains stretches of ambient
noise that you can use to judge the relative loudness of the subject. Be aware
that very quiet subjects may not be audible — if we couldn’t hear it from
one meter, chances are we couldn’t record it either!
The recording starts with 5~10 seconds of room ambiance, followed by 5~10 seconds
of the VGA test system without a video card installed, and then the actual product’s
noise at various levels. As this particular card did not add any noise the test
system, we have provided only a recording of the test system with its system
fan set to the levels tested. For the most realistic results, set the
volume so that the starting ambient level is just barely audible, then don’t
change the volume setting again.
Gaming: Please check out gaming-oriented reviews of the HD 4890 at techPowerUp,
Tech Report. The general consensus is that the HD 4890 is about 15% faster
than the 1GB version of the HD 4870, a nice performance bump considering the
price (especially since ATI’s price cut was never made reality). Overclocked
versions like the HIS Turbo of course will perform slightly better but come
at a premium. It is an ideal graphics card for playing most modern titles at
Cooling: The stock cooler is plainly audible when idle and its fan
ramps up very quickly when the card is stressed. It does deliver pretty good
temperatures, but the cost in noise is much too high. We’ve never been a huge
proponent of ATI’s stock coolers, but at least they kept the card practically
inaudible when idle and at reasonable levels when stressed. The speakers would
have to be cranked way up during gameplay to drown out the 4890’s fan. The aggressive
fan behavior has been noted on various 4890s by other hardware sites, so it
does not seem to be specific to our sample or even HIS’ version.
Power Consumption: By our estimates, the HIS Radeon HD 4890 Turbo requires
about 71W idle and up to 149W when stressed to the limit. This is an increase
of 4W idle and 15W load compared to the 4870, not a big difference when you
consider the extra performance. That being said, ATI either needs to work some
serious mojo on their PowerPlay technology or move on to a smaller manufacturing
process if they wish to get their idle power usage down to nVidia’s level.
The HD 4890 offers a moderate performance boost over the HD
4870 and merits significant consideration if you are looking for a $250
graphics card. The HIS Radeon HD 4890 Turbo is currently retailing for $300,
but other overclocked versions can be had for a bit less. If noise
is a concern, keep in mind that the aggressive fan control of cards utilizing the reference cooler
makes it substantially louder than the 4870 cooler. nVidia also has a new card
to compete directly against the 4890, the GeForce GTX 275; No comment beyond that here until we get one to examine closely.
HIS Radeon HD 4890 Turbo Edition
* Excellent 3D performance
* Fan aggressive, too loud
Our thanks to HIS
for the video card sample.
* * *
Articles of Related Interest
GeForce GTS 250 1GB Graphics Card
EN9400GT Silent Edition
Radeon HD 4870 Matrix
PowerColor SCS HD4650: A Fanless Budget Graphics Card
EN9800GT Matrix Edition
the Gap: ATI Radeon HD 4830
* * *