What's new
  • Please do not post any links until you have 3 posts as they will automatically be rejected to prevent SPAM. Many words are also blocked due to being used in SPAM Messages. Thanks!

ATI Radeon HD 4890 Roundup (ASUS, Diamond, HIS, Sapphire, XFX)

Status
Not open for further replies.

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Tom Clancy’s Hawx (DX9)

Tom Clancy’s Hawx (DX9)


HD4890-26.jpg

Even though HawX has a built-in benchmarking tool, we found that it did not provide accurate enough results needed to reflect what actual in-game performance would be. Thus, we played through the first 6 minutes of a mission and recorded the results with FRAPS.

1680 x 1050

HD4890-ROUNDUP-76.JPG


HD4890-ROUNDUP-77.JPG


1920 x 1200

HD4890-ROUNDUP-78.JPG


HD4890-ROUNDUP-79.JPG


2560 x 1600

HD4890-ROUNDUP-80.JPG


HD4890-ROUNDUP-81.JPG
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Tom Clancy’s Hawx (DX10 / 10.1)

Tom Clancy’s Hawx (DX10 / 10.1)


HD4890-26.jpg

Even though HawX has a built-in benchmarking tool, we found that it did not provide accurate enough results needed to reflect what actual in-game performance would be. Thus, we played through the first 6 minutes of a mission and recorded the results with FRAPS. DX10.1 was enabled for ATI cards.

1680 x 1050

HD4890-ROUNDUP-82.JPG


HD4890-ROUNDUP-83.JPG


1920 x 1200

HD4890-ROUNDUP-84.JPG


HD4890-ROUNDUP-85.JPG


2560 x 1600

HD4890-ROUNDUP-86.JPG


HD4890-ROUNDUP-87.JPG
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Heat & Acoustics

Heat & Acoustics


For all temperature testing, the cards were placed on an open test bench with a single 120mm 1200RPM fan placed ~8” away from the heatsink. The ambient temperature was kept at a constant 22°C (+/- 0.5°C). If the ambient temperatures rose above 23°C at any time throughout the test, all benchmarking was stopped. For this test we use the 3DMark Batch Size test at it highest triangle count with 4xAA and 16xAF enabled and looped it for one hour to determine the peak load temperature as measured by GPU-Z.

For Idle tests, we let the system idle at the Vista desktop for 15 minutes and recorded the peak temperature.


HD4890-ROUNDUP-96.JPG

Overall, the HD 4890 is a loud card. Some manufacturers (Diamond and HIS) represented here today took it upon themselves to add a bit less aggressive fan speed profile which resulted in slightly higher temperatures but even these minor changes didn’t help things much. The stock cooler on the HD 4890 relies on brute force fan speed to keep its temperatures within check rather than a well designed heatsink assembly. This is a shame since many of us value peace and quiet when we are gaming and in this, the newest ATI card is an absolute failure. We recommend setting the fan speed to a constant 40% and living with higher temperatures if noise is an issue for you.

Sapphire on the other hand has harnessed the power of their Vapor-X technology to design a heatsink that manages to cool off the highest-clocked core of the group to downright impressive levels. Add to this the fact that the fan is barely a whisper over the noise of even a Yate Loon 120mm 1200RPM case fan and we are definitely happy that one board partner could put two and two together.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Power Consumption

Power Consumption


For this test we hooked up our power supply to a UPM power meter that will log the power consumption of the whole system twice every second. In order to stress the GPU as much as possible we once again use the Batch Render test in 3DMark06 and let it run for 30 minutes to determine the peak power consumption while letting the card sit at a stable Windows desktop for 30 minutes to determine the peak idle power consumption. We have also included several other tests as well. Please note that after extensive testing, we have found that simply plugging in a power meter to a wall outlet or UPS will NOT give you accurate power consumption numbers due to slight changes in the input voltage. Thus we use a Tripp-Lite 1800W line conditioner between the 120V outlet and the power meter.

HD4890-ROUNDUP-94.JPG

The law of power consumption and graphics cards usually goes something like this: increase clocks and more power is needed. Nearly all of the cards featured in this roundup adhere to this law…all except one. Somehow, Sapphire managed to keep power consumption below that of lower clocked cards. We don’t know how they did this but after testing over and over again, the same facts presented themselves. Unfortunately, Sapphire couldn’t help with the abysmal idle power consumption that seems to plague the HD 4890 series.
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Overclocking Results

Overclocking Results


What do we consider a stable overclock? If the card can pass our test which involves a 20 minute loop of 3DMark06, 30 minutes of playing Far Cry 2 and an additional 30 minute gameplay sequence in Left 4 Dead, it is considered stable.

Since overclocking can vary from one card to the next within the same family of cards, you should be aware that these results will not be considered when voicing our final opinions about these four cards. As such, we will show you the final overclocking results but leave our opinions to ourselves and let you guys form your own conclusions.

A Note About Voltages

Please note that the default core voltages on the cards are as follows:

Stock: 1.312V
ASUS: 1.312V (Adjustable up to 1.45V)
Diamond: 1.312V
HIS: 1.312V
Sapphire: 1.36V
XFX: 1.312V

At the time of writing, the voltage used for the overclock on the ASUS TOP Edition was 1.37V through Smartdoctor. We are still experimenting with additional voltages and will update the overclocks in the comment thread in our forums.

In addition, the HIS Turbo was excluded from the overclocking section due to the fact we found one that ran the proper clock speeds only two days before this roundup went live. As such, we did not have the time for overclocking stability testing.


ASUS HD 4890 TOP

Final Overclocks

Core: 1026Mhz
Memory (DDR): 4560Mhz

HD4890-ROUNDUP-407.JPG


Diamond HD 4890 XOC

Final Overclocks

Core: 968Mhz
Memory (DDR): 4644Mhz

HD4890-ROUNDUP-405.JPG


Sapphire HD 4890 Atomic

Final Overclocks

Core: 1035Mhz
Memory (DDR): 4420Mhz

HD4890-ROUNDUP-403.JPG


XFX HD 4890

Final Overclocks

Core: 1004Mhz
Memory (DDR): 4784Mhz

HD4890-ROUNDUP-401.JPG
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Conclusions (ASUS & Diamond)

Conclusions


Before we actually get to the individual conclusions, we really have to talk about the HD 4890 in general and how well it has been shaping up over these last few months. In our original review of the card, we mentioned that the only way ATI would succeed with this card would be to have a hard launch with product in the channel and ready to buy on launch day. They definitely did just that and even Nvidia’s paper launched GTX 275 could do little to take the wind out of ATI’s sails.

The stock situation with the HD 4890 cards has remained very good but I feel that the real story here is just how scalable ATI’s current architecture really is. Back when the HD 4870 was first launched, ATI made it apparent to us that they weren’t necessarily going for performance domination but rather for smaller chips that could be adapted for numerous situations. Remember, the chip that was premiered in the HD 4870 512MB has come a long way and barring a few tweaks, is now –nearly a year later- powering the beast that is the HD 4890. If that isn’t proof that ATI’s vision has come true, I don’t know what is.

With new cards approaching, we are seeing the twilight of the current architectures and let me tell you; the HD 4890 is one hell of a swan song for this generation of ATI cards.


ASUS HD 4890 TOP

It seems that no matter what happens in the industry, one thing will always be constant: that ASUS will release a high quality product time and again. Their HD 4890 1GB TOP Edition is just the latest in a long line of ASUS cards that have put their best foot forward and have impressed us as a result.

Even though the higher clock speeds made this card’s performance look great on paper, I can tell you now that in most regular gaming scenarios you won’t be able to see a performance difference between the TOP and a stock-clocked HD 4890. Granted, there will be some instances where minimum framerates will hold steady above 30 where the stock card falters a bit. But are those few and far between cases worth this card’s $50 price premium over most reference-based cards? We don’t think so.

What does make the ASUS TOP unique is the Voltage Tweak feature. Usually we don’t add overclocking observations into a conclusion but with the ability to increase voltage up to 1.45V through SmartDoctor, this card can really fly. Just remember, with increased voltages come increased heat output so make sure you have sufficient cooling.



Diamond HD 4890 1GB XOC

When Diamond first approached us with the opportunity to review their HD 4890 XOC, we had some very high expectations. The clock speeds looked amazing since at the time it was the highest clocked HD 4890 on the market and the price definitely looked right. Add to that the fact that Diamond was officially doubling their warranty length from the original single year (which was well below industry norms) to two years and I found myself really hoping for something special. Well, they delivered.

Not only was the XOC able to run away from the much more expensive GTX 285 1GB on a regular basis but due to its high memory clocks, its performance sat right behind the ultra high speed Sapphire HD 4890 Atomic. That in itself is impressive but unlike the ASUS card featured here, Diamond clocked this card high enough that the difference between it and a stock clocked card is perceptible in nearly every game. The minimum framerate increases we saw were impressive to say the least.

I can talk all I want about the performance of the XOC but that wasn’t what impressed me the most about it. Rather, it was the price that really had me stand up and take notice. What if I told you that at the time of writing, this little pocket rocket was retailing for about $12 higher than a stock HD 4890? Would you be interested? I know I would be which is why Diamond’s HD 4890 1GB XOC receives our Dam Good Value Award.

24046d2ee78a9861.gif
 

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Conclusions (HIS & Sapphire)

Conclusions



HIS HD 4890 1GB Turbo

Ah HIS, what can we say about thee? Based on its price / performance ratio alone, their HD 4890 Turbo had some real potential to be one of the forerunners in this roundup. It was able to offer performance above and beyond that of a stock HD 4890 while retailing for approximately the same price and availability is pretty good as well. That my friends, is a recipe for success.

Unfortunately, an apparent BIOS flashing mishap seems to have left many of their supposedly overclocked cards running at stock speeds. As we said several times in this roundup: after spending over $1000 for five HIS Turbo cards from several retailers, only one of them ended up having the correct clock speeds. Luckily, because of that lone card we were able to include the Turbo into our roundup. We don’t even know if these issues are limited to the Canadian market for the simple reason that HIS has yet to return any of our emails over the past two weeks. If this is how they treat all their potential customers, then shame on them.

We need to hear from HIS regarding how far this issue extends into the retail channel and what kind of measures will be put in place to take care of customers who paid for an overclocked card but received one with stock clocks. Until then, we are going to do something we have never done before: recommend you avoid the HIS HD 4890 Turbo at all costs.

If you have this card, we invite you to post in this thread on our forums so we can get some idea if these issues are limited to the Canadian market or are worldwide.



Sapphire HD 4890 1GB Atomic

In the past I have talked about how reviewing products on a daily basis will cause a reviewer’s opinion of generic hardware to become a bit jaded. As such, it takes a hell of a lot for me to sit up and get really worked up about a product but the Sapphire HD 4870 Atomic had me beside myself with excitement. I couldn’t wait for benchmarks to finish so I could see what insane performance levels this card would post next. At times it felt like I was dog running circles around a family picnic waiting for the next bone to be thrown on the ground.

Performance is one thing but the Atomic gets so many things right that it is impossible to find any faults with it. It will be priced at around $300 USD which makes it quite a bit less expensive than a GTX 285 and even many GTX 275s on the market. It also has an accessory package filled with useful applications and cables. Even its heatsink performs miracles with a core that is clocked far above many of the cards featured here but goes about its job in near silence. In all reality, the only issue I really have with this card is that availability is non-existent.

Every now and then, the planets align and the Hardware Gods grace us with that absolutely perfect piece of hardware. Well, the Sapphire HD 4890 Atomic is just that and more. We have never done this before but I’m going to give this card every award we’ve got. The Dam Good Award because it really is that good. The Dam Innovative Award due to its lower than expected power consumption, low noise output and amazing temperatures. Finally it gets the Dam Good Value Award because it more often than not grabs a stock GTX 285 and smacks it around while yelling “who’s your DADDY!?!” over and over again while retailing for about $50 less. Bravo Sapphire.

24046d2ee78a9861.gif
240463923aca1f6b.jpg
dam_innovative.png
 
Last edited:

SKYMTL

HardwareCanuck Review Editor
Staff member
Joined
Feb 26, 2007
Messages
12,840
Location
Montreal
Conclusions (XFX & Final Thoughts)

Conclusions



XFX HD 4890 1GB XT

Many of ATI’s longtime faithful have had been waiting for a company to come along and offer them the same type or warranty length and support that had been offered to Nvidia’s customers for years. With XFX now an official ATI board partner, their wishes have come true. Not only did they carry over their excellent customer support but their unique Double Lifetime Warranty is also offered on all Radeon cards. A winning combination likethis means that XFX products are some of the best selling ATI-based SKUs on the market today.

The HD 4890 XT is a bit of an unassuming card considering its insignificant overclock is really nothing more than window dressing in the grand scheme of things. Its performance is fractionally above that of a reference HD 4890 but in all reality, a customer will never be able to tell the difference.

Regardless of performance in comparison to higher clocked cards, where the HD 4890 XT succeeds is in bringing some serious value to the table. XFX’s double lifetime warranty alone brings their customer’s piece of mind that none of the other board partners in this roundup can lay claim to and for some people it will prove to be an invaluable selling point. In addition, the inclusion of a full version of HawX just adds the icing on the cake. I am sure you know where this is already going but let’s make it official nonetheless: the XFX HD 4890 XT wins our Dam Good Value Award.

24046d2ee78a9861.gif



Final Thoughts

I am sure there are many of you who looked at our charts and saw the EVGA GTX 275 FTW and Sapphire HD 4890 Atomic battling it out for the pole position again and again. Is there really a winner here? Personally, I don’t think so and let me tell you why. I have a basic law of video card performance which states that Nvidia cards will perform better in some games while ATI products will perform best in others. It is no less true than saying water flows downhill or that we will all eventually die one day. Consequently, in traditionally Nvidia-centric titles the FTW walks away with things and not even the 1Ghz core speed of the Atomic can change anything about that. Flip that argument around and it is no less true. What we can say is that in some cases, ATI’s gamble with DX10.1 seems to be paying dividends.

Another question that has been asked has been this: what does it take for a HD 4890 to beat a stock GTX 285 in the majority of the benchmarks? You need the speeds of the Diamond HD 4890 XOC at lest and even then, it is a close thing. That being said, the HD 4890 Atomic definitely does the job of taking the GTX 285 to task if you can find one.



Thanks to all the companies who participated in this article​

 
Last edited by a moderator:
Status
Not open for further replies.

Latest posts

Top