In this article, I’m comparing the Palit 9500GT 256MB DDR3 and Palit HD4650 Super 512MB DDR2. DDR3 vs DDR2? 256MB vs 512MB? Yes, I’ve decided to compare these cards as they are of similar price range.
The Appearance
They come in tiny neat looking boxes. With the RED vs BLUE them, it almost feels like The Matrix where Neo chooses between the RED and the BLUE pill.
I was expecting the nVidia card to be something that’s NOT RED. To me, RED has always been the ATI board while I’ve encountered a few RED nVidia boards, I certainly was really expecting Palit to give this card a blue PCB to match the box. 🙂
The cards are very simple looking. In fact they looked almost the same, it’s just that the HD4650 unit has a bigger heatsink and occupies 2 slots.
Card Details & Bundle
Both cards come with VGA, DVI and HDMI out. Neat!
What I’m most puzzled is that my 9500GT unit doesn’t look anything at all like the one on the 9500GT product page. Can’t find anything on the product list that looks anything like it. So I’m paying for the same thing minus the better heatsink?
At least for the HD4650 unit, what I see is what I got – exactly as what it is in the HD4650 Super product page.
Here’s a summarized detail of the card.
Model | 9500GT | HD4650 Super |
GPU | 9500GT | HD4650 |
GPU Code Name | G96 | RV730 |
Memory Type | DDR3 | DDR2 |
Memory Bandwidth | 128-bit | 128-bit |
Memory Capacity | 256MB | 512MB |
Core Clock | 550Mhz | 600Mhz |
Memory Clock | 800Mhz (1.6Ghz DDR) | 400Mhz (0.8Ghz DDR) |
Both units come with Manual and CD only, the 9500GT came with an additional cable but nothing in the box says what’s it for.
Test Setup
The computer setup used for this review.
Processor | Intel Pentiu Dual-Core E2140 @ 2.4Ghz |
RAM | 2GB OCZ DDR2 PC2-8500 Reaper HPC Edition |
Motherboard | Albatron PXP35 |
Cooler | Stock Heatsink |
Monitor | 20″ Dell Ultra-sharp Wide-screen LCD |
Operating System | Windows Vista Home Basic 32bit |
Driver Version | nVidia GeForce 181.22 and ATI Catalyst 9.1 |
The Performance
Seeing that both are low end cards, I’ve decided to test them on 1024×768 resolution as well. Just to see how far has technology brought us, even on rather old games.
F.E.A.R.
Graphic Detail : Maximum, 4xAA
Card | 1680×1050 | 1024×768 |
9500GT | 23 | 54 |
HD4650 | 17 | 23 |
World In Conflict
Graphic Detail : Maximum, 4xAA
Card | 1680×1050 | 1024×768 |
9500GT | 4 | 8 |
HD4650 | 6 | 10 |
Company Of Heroes : Opposing Fronts
Graphic Detail : Maximum, 4xAA
Card | 1680×1050 | 1024×768 |
9500GT | 7.6 | 19.5 |
HD4650 | 15.3 | 27.5 |
Lost Planet : Extreme Condition
Graphic Detail : Maximum, 4xAA.
Card | Snow (1680×946) | Cave (1680×946) |
9500GT | 12.6 | 18.6 |
HD4650 | 10.0 | 13.5 |
Card | Snow (1152×648) | Cave (1152×648) |
9500GT | 23.6 | 33.7 |
HD4650 | 18.4 | 23.8 |
Crysis
Graphic Detail : Maximum (Very High) 4xAA
Card | 1680×1050 | 1024×768 |
9500GT | <1 | 6.96 |
HD4650 | 4.04 | 8.79 |
(Yes! no typo there! It is <1! Even the MENU screen took minutes to load. Those red boxed notices took around a minute to load. A MINUTE! ALT+TAB, Task Manager to exit game was the fastest out!)
Furmark (OpenGL)
Graphic Detail : 4xAA
Card | 1680×1050 | 1024×768 |
9500GT | 17 | 5 |
HD4650 | 9 | 5 |
Confused? Looks like both cards exchange blows for the benchmarks. In the end, it’s the card with the faster memory that’s the overall winner. The HD4650 with twice the memory size does lead in games like Crysis and World in Conflict but the lead isn’t much, considering both are unplayable.
I know I’m testing these cards pretty harshly but I think it’s a fair test to see whether they could cope with maximum details even at 1024×768.
Generally, you can see that both cards work best at lower resolution and in some cases, much lower detail is required. Like say disabling Anti-Aliasing completely and reducing details on textures and models then both cards would run the games nicely, and perhaps you could possibly bump it up to 1440×900 for 19″ wide LCD if you’re fine at running lower details.
Operating Temperature
Great temperatures.
Card | Idle °C | Load °C |
9500GT | 44 | 56 |
HD4650 | 39 | 52 |
The HD4650 has the advantage with the bigger heatsink, more surface area for heat dissipation.
Conclusion
I have no problem giving both cards the recommendation…….
BUT among the 2, I personally would go for the 9500GT over the HD4650 as it’s generally faster (other than the fail in Crysis 1680×1050 test).
Make no mistake, the HD4650 DDR3 version should certainly be a lot faster but good luck finding it among the distributors in Malaysia. From what I gathered, most distributors don’t carry the HD4650 DDR3 version in order to avoid killing the other products. Seriously, if you’ve seen the benchmarks of the HD4650 DDR3 version, it’s a much faster card than 9500GT but as for the DDR2 version, it’s crippled.
Having the DDR3 version will hurt either the DDR2 version or the HD4670 sales as the performance will be in between the HD4650 DDR2 and HD4670 DDR3.
I was expecting that the hd 4650 512mb ddr2 is faster and was planning to buy one good thing I saw your reviews and benchmark
🙂 Good, usually stuff with DDR2 don’t perform as well as those with DDR3.
Why do you test with all settings cranked up? What good is 4xAA if the cards are delivering completely unplayable framerates? More often than not, the framerate drops below 20.
It makes no sense to call a winner with these benchmarks IMHO, some cards degrade faster in terms of fps when cranking up the graphics than others.
To put things into correct perspective – that is why. 🙂
See, what’s the point if run it at low detail test so that the results show very playable framerates? It’s not showing the reality of things, some people are are dumb enough to see the results yet not referring to the settings, so they concluded thinking the cards were fast then later when things turn out the other way, they start blaming others.
I tested it on high resolution and low resolution (1024×768) and also to at the usual full settings, even with 4xAA, to make it absolutely clear that people would understand that these are low end cards and that they should have the right understanding.
And just as what I highlighted in the paragraph before Operating Temperatures, I stated …….
“Generally, you can see that both cards work best at lower resolution and in some cases, much lower detail is required. Like say disabling Anti-Aliasing completely and reducing details on textures and models then both cards would run the games nicely, and perhaps you could possibly bump it up to 1440×900 for 19″ wide LCD if you’re fine at running lower details.”
As you can see, even with full details it’s still playable on some games. Games that are older.
i agree with Stormcrow,
your benches provide no decision making value for potential buyers. people who are looking at these cards obviously do not require 4x AA and highest quality settings in games.
it would be more useful if you could run the benches on medium graphic details and without AA. at least then, users could decide which card would be more suitable for them to use at a particular resolution with fair quality settings.
if you refer to reviews on the web on lower end graphic cards, you will notice that these reviews focus on providing reasonable settings for users during the benches.
of course, being the great debater you are, you could argue that in the end, the benches still do prove 9500GT as the overall winner, but please keep in mind that the user still has no idea on what type of performance the cards can deliver on usable settings.
to put it into a simpler conclusion. the users would still have no idea on whether or not the cards could provide a decent gaming experience at a particular resolution due to the benches that provide no real meaning to them.
writing a review to prove stupid people wrong rather than to provide information to people who want to know. great logic in that!
@anon, i really appreciate the feedback.
there’s a reason why i had a lower resolution test (1024×768) which i think it shows clearly that both the cards work great at lower resolution, even with max detail. on the other hand in some games, lowering the resolution didn’t help so people will have to settle with lower resolution on it.
looking at the low resolution test, i’m sure it’s quite clear that these cards work well even on HIGH settings with AA on some games, other than Crysis and World in Conflict – i think is more than sufficient to understand the card. furthermore there’s the comment section here for people to ask. 😀
however, you have good and valid points, they’re noted. I’ll provide benchmark for PLAYABLE settings for high resolutions next time.
@anon2 – erm. thanks for the sarcasm. 😀
there is better the 9500GT because is DDR3, that’s the real signify of a graphics card acelerattor. Do you understand? And people who haven’t got an [WIDESCREEN] with 1024×768 are fine. I’m sorry by my bad english, because i’m Argentinian and i don’t know too much english but yes everything about PC.
well my advice is 9500GT vs HD4670 are same the only differences of this two here and the reviews is HD4670 has DDR2 while 9500GT has DDR3 which has the advantages did you see how close they are? if HD4670 has DDR3 then it well probably PWND 9500GT , the Memory Capacity is not option here it doesn’t matter so for sure and HD4670 is greatly faster card i would recommend this card for heavy games like Crysis on Very High with 30+ FPS
Eh, it’s HD4650. 🙂
There is a HD4650 DDR3 version but it was rare and I think it still is, simply because the HD4650 DDR3 version would probably kill off the sale of cards like HD4670.
At the time of the review, only 9500GT and HD4650 DDR2 available and they were compared here due to similar pricing.
i would agree with most people here
am building an cheap HTPC and need some more info on exact settings to get +30 framerate on those games mentioned.
i mean the review does give good info and proves a point, but not the details to get +30fps (which is what we all want to know)
cheers 🙂