Overclocking
Now on to the fun part of the review, overclocking the video card. HIS makes the claim that the card is already slightly over the stock settings, which they did by raising the core 25MHz. Lets see what kind of cooling this card offers and how far it will overclock.
We can see as is pretty normal for HIS, the cooler is very large but surprisingly not heavy at all, unlike the copper heatsink on the Asus x800XL we compared it to. The ram-sinks are held on by thermal tape so we couldn't remove them very easily.
Looking at the overclocking results I was a little disappointed, as the core only went up another 33MHz and the memory 60MHz(120DDR). This represents a ~5% increase in the core clock and a ~10% increase in the memory clock speed. One very nice thing though was the temperatures even at this increased speed. The core never went over 56ºC which is quite a nice result as many cards routinely reach the 70ºC and higher mark under load. The card is also very quiet in use, as I didn't realize the computer was on in my new Athlon 64 board until I saw the fan spinning.
System Setup
| CPU: |
Intel Pentium IV 520 2.8E (14*200MHz)
|
| Motherboard: |
Asus P5LD2-Deluxe
|
| Memory: |
2*512MB Crucial PC2-6400, 2*512MB SyncMax 533Express |
| Hard Drives: |
40GB Maxtor IDE, 5*250GB WD SATA (RAID5)
|
| Video Card: |
HIS X1650Pro IceQ Turbo Dual DL-DVI - 625/700MHz - 658/760MHz |
|
Asus Extreme AX800XL-2DTV - 398/490MHz |
| Operating System: |
Windows XP Professional Service Pack 2 w/Direct X 9.0c |
| Drivers: |
HIS ATi Catalyst 6.7 and ATi Catalyst 6.8 |
| Cooler: |
Zalman CNPS7700-AlCu |
| Case: |
Generic ATX Case |
| Power Supply: |
Ultra X-Finity 500 Watt
|
| Software: |
Fraps 2.5.3 |
|
| Direct X Benchmarks: |
Serious Sam 2 Demo - Self Demo |
FEAR - Built In Demo |
|
Half Life 2: Lost Coast - |
|
| OpenGL Benchmarks: |
Quake 4 1.3- Self Demo |
|
All our tests were run at 1024*768 and 1600*1200 with 4x AA and 16X anisotropic filtering both enabled and disabled. The x1650 wasn't detected by the Catalyst 6.8 drivers so we used the 6.7 drivers for its testing while using the 6.8 drivers for the x800XL. The overclocked version of the x1650 was only tested in the more demanding settings, 4x AA and 16X anisotropic filtering at both resolutions.
The Quake 4 demo was recorded in the first level, as you leave for initial area, and covers both outdoor fighting as well as indoor areas. The F.E.A.R demo was the built in performance demo included with the game, and the basic settings are shown here. Serious Sam 2 Demo was recorded just after the initial staging area of the demo, in a mostly closed in area with many enemies as is standard for Serious Sam, the settings used are located here. Half Life 2: Lost Coast was used as I don't have Episode 1, but this is pretty close. The demo used is from Tech Report, and covers most of the initial area in the map, and we used the settings shown here. So lets see what this new mid-range card can do.
Game Tests - Quake 4
First lets take a look at our only OpenGL based game, in Quake 4. Based off the Doom 3 engine this game has some definite high quality graphics, or we wouldn't use it for testing, so lets see what the results are:
|
1024 NAA - Results (In FPS)
|
Minimum |
Average |
Maximum |
| Asus Extreme AX800XL: |
29 |
52.5 |
94 |
| HIS x1650 Pro: |
36 |
57.59 |
136 |
We can see that the HIS card does better than the Asus card in this test. Most of the difference comes in the first four seconds of the test, where the HIS card goes to its maximum of 136fps while the x800 has a 40fps lower maximum, which alone is a 1fps difference in the average frame rate. Both cards can be considered very playable in this resolution as they don't go under 35fps apart from the initial 29fps from the x800XL. Lets turn up the quality settings a to see if there is a wider difference between the cards.
|
1024 AA - Results (In FPS)
|
Minimum |
Average |
Maximum |
| Asus Extreme AX800XL: |
33 |
54.69 |
122 |
| HIS x1650 Pro: |
35 |
56.57 |
126 |
| HIS x1650 Pro 658/760MHz:: |
34 |
56.93 |
128 |
Here we can see what appears to be a CPU limited result for the most part. Strangely enough the average results are higher than the tests without AA and anisotropic filtering at this resolution. Again all three cards are playable with frame rates under 40fps only occurring for 2-4 seconds per card in the entire test. The graphs are fairly identical during the second half of the test, which equates to the outdoor portion of the test, thus most of the differences occur in a more confined rendering area. Lets see if the past two results continue with the resolution bumped up to 1600*1200.
|
1600 NAA - Results (In FPS)
|
Minimum |
Average |
Maximum |
| Asus Extreme AX800XL: |
36 |
55.14 |
127 |
| HIS x1650 Pro: |
27 |
51.70 |
104 |
The tables seem to have changed from the 1024*768 results, as the x800XL is in top spot by a 4fps margin. This could be a result of higher fillrate and memory bandwidth, as the x800 has a 256bit bus compared to the 128bit bus of the x1650 series. However besides this the cards both are playable at this setting, which is pretty good for a mid-range card. Lets see if adding some enhancements will bring both cards to their knees.
|
1600 AA - Results (In FPS)
|
Minimum |
Average |
Maximum |
| Asus Extreme AX800XL: |
23 |
42.8 |
95 |
| HIS x1650 Pro: |
24 |
45.02 |
78 |
| HIS x1650 Pro 658/760MHz:: |
22 |
46.52 |
78 |
Again just like at 1024*768 the results are very close, the only difference is the average frame rate has dipped about 10fps on average between the cards. None of the cards dips below the 20fps mark at all and for most people they should find this game playable at this resolution, which is a very nice thing to be able to say about any mid-range card. But before we hand out any medals, lets see how it does in the newer DirectX 9c (PS3/VS3) based games.
NEXT
