Sapphire 4890 Vapor-X PCIe Graphics Card

Packaging & Appearance

Packaging & Appearance
Straight out of the blocks, the Sapphire 4890 Vapor-X looks ice cold to the touch with Antarctic scene. I must say this is a pleasant break from the tired routine of displaying CGI Characters and as aesthetically pleasing ATI's Ruby might be, I do tire of seeing her plastered on every ATI product I come across. It's plainly obvious by the front of the box that Sapphire intend to promote this cards cooler being that the Vapor-X emblem takes pride of place in the centre of the box. A brief run down of the features are also displayed on the front with the main title of Radeon HD4890 displayed towards the bottom of the package on an icy silver backdrop. Flipping the box over, Sapphire go on to display the product highlights and features in greater detail.
box front box rear
One thing I have heavily criticised Sapphire for in the past is the lacklustre packaging methods but the Vapor-X 4890 has seen a revised method of using a plastic insert to hold the anti-static, bubble wrapped card in place. Beneath this is a further cardboard box holding the accessories meaning a much improved package overall. The included accessories quite good with a DVI-HDMI dongle, two twin Molex to PCI-6 pin power cables, a Crossfire bridge, PowerDVD, DVD suite, a copy of Futuremark Vantage and the driver CD concluding the list.
inner box accessories
The card itself is a very attractive affair with the Vapor-X cooler sitting atop a fetching blue PCB. While I'm not a great fan of coloured PCB's, blue does match the cards aesthetics quite nicely. The card is as long as a standard ATX motherboard is wide so there should be no issues with fitting the card in a standard ATX sized case. A further heatsink sits abreast the MOSFETs with the Sapphire name emblazoned across the top.
card top card bottom
The black plastic fan is very quiet in operation being almost silent when idle. Cool air is taken in from the fan and then blown over the Aluminium finned heatsink underneath which in turn cools the copper Vapor chamber. Because of the thickness of the cooling design, the card is a dual slot version, much the same as the reference cards. Like the reference ATI cooler, the Vapor-X cooler is partially shrouded to prevent hot exhaust air from escaping into the PC case and instead directs the flow to the rear finned I/O plate.
fan card side
The I/O plate itself interestingly houses DVI, VGA as well as HDMI ports allowing all the technology of today to be connected through this card. At the back end of the card we see that the 4890 requires 2x 6-pin PCI power cables to be connected despite the cards revised power circuitry. Regulating the power behind the PCIe ports are solid capacitors ensuring greater reliability of the card along with Sapphires own 'Black Diamond'  power choke which Sapphire claim run 10% cooler and offer up to 25% better power efficiency than a standard choke.
PCI card rear
Removing the heatsink assembly was as easy as removing a stock reference heatsink as just four screws hold it in place around the GPU core area. A further two screws hold the MOSFET cooler in place. With these coolers removed we can see the 4890's nakedness in all it's glory.
card naked card base
The Vapor-X cooler is based on the same principles of heatpipe technology we have become familiar with in the past with a liquid coolant vaporising closest to the core and then condensing towards the top of the cooler only to be returned back to the hot surface. If you want to know more about Vapor-X click, HERE. How good this process works will be examined in our test setup area when we place the card under some extreme load conditions. Nevertheless, it is nice to see Sapphire have used copper as the contact surface rather than inefficient aluminium.
contact core
The Sapphire HD4890 Vapor-X uses the same memory chips as the stock reference cards. In this case however the 1GB of GDDR5 Qimonda chips are clocked to a massive 4.2GHz. This will hopefully negate the very conservative GPU core overclock. Below right, we see the 16 MOSFET's which also interestingly have space for an additional two. This leads me to believe that perhaps 14 are used for the core with two reserved for the memory being that this version is the 1GB GDDR5 rather than the separately available 2GB which would likely incorporate the additional MOSFETS required to regulate the power of an extra 1GB.
memory fets
Sapphire have designed a very attractive card, not only in the packaging but also the card itself. The accessory list is complete enough to get everyone started and one might even wish to venture into the world of benchmarking given that Sapphire have also included Futuremarks Vantage. I would perhaps liked to have seen an included game but if this omission keeps the price down then I can see the reasoning behind not including one.
Let's take a look at the test setup I will be using today to evaluate the performance of the Sapphire HD4890 Vapor-X... 
«Prev 1 2 3 4 5 6 7 Next»

Most Recent Comments

02-06-2009, 06:21:08

"Can Sapphire's vapor X cooler keep this maniacal overclocker of a card in check or is it just a lot of media hype and hot wind? Find out in our latest GPU review..."

Sapphire 4890 Vapor-X PCIe Graphics CardQuote

02-06-2009, 06:42:45

the card peforms extremely well, but the card looks horrible when will sapphire discover good looking card chassisQuote

02-06-2009, 07:29:13

Sweet review man, massive props to Sapphire for final sticking a non reference cooled card on the market. Any idea when these will be available?Quote

02-06-2009, 14:51:16

Nice review as always my good man!

I wasn't sure that these coolers would make much odds over other after market coolers but those temps are pretty damn good.Quote

08-06-2009, 12:40:40

very nice review!

Is that cooler good enought?Quote

30-06-2009, 18:05:52

Well actually I am not quite convinced. You see almost solely based on this review I went merrily our and promptly bought 2 of these cards only to find that .

1) They run a lot hotter than stated. 67 Celcius at idle and run so hot when gaming that my PC crashes. I haven't managed to record the temp just before crash-poin but I have seen it at upto a blistering 93 Celcius!

2) I don't get anywhere near the FPS stated in the article. Infact I get exactly the same when playing CoD World at War online with 2 of these cards in crossfire as I did with one 9800GTX+. Namely 91 FPS.

I have plenty of ventilation in my rig and the fans on the HD4890s are certainly working.

At +4000rpm (confirmed by ATIs own CCC program and CPUIDs Hardware Monitor) they are audible let me tell you.

I don't believe my system is to blame despite it not being cutting edge.

Sapphire recommend a bigger PSU. 650W-750W and monitoring my temps with rivatuner.

I have a 620W PSU and run Vista ultimate 64 bit, which doesn't allow rivatuner to work since I downloaded the latest update if you can call it that.

Sys Specs:- Q9550@3.2GHz with Thermalright Ultra-120 eXtreme cooler and Noctua NF-P12 fan. ASUS Maximus Formula Republic of Gamers motherboard with X38 chipset and 4 GHZ of Kingston HyperX PC9600 RAM @ 1207MHz being cooled with a Corsair Dominator Airflow Fan.

CoolerMaster Real Power M620 Modular PSU.

Vista Ultimate 64 bit OS.

Samsung SpinPoint T166 HD501LJ 500 GB.

All incased in a Aerocool Hi-Tech 7 cabinette with a 40cm fan on the side and 2 x 12cm fans, one infront and one at the back.

Any good advice would be appriciated.

Bilko out.Quote

01-07-2009, 18:49:20

Hey Bilko, believe it or not your system is a quite a lot inferior than our test system.

You CPU is not only slower than our i7 CPU but it is also clocked a lot slower. CPU bottlenecking, which you will almost certainly have with 2 4890's at your current CPU & clockspeed, is the most likely culprit in why you are not seeing good FPS. You are also running with 4GB of DDR2 at a much slower speed than the test systems 6GB of DDR3 which is running 400MHz faster than yours. I don't wish to put your system down as it sounds like a sweet setup but our test setup is specifically designed not to bottleneck any product we throw at it and as such any product we review is done so under optimum conditions.

Rivatuner works fine with Vista 64bit. I use it to overclock NVidia cards.

Your PC is most likely resetting due to power requirements I'm afraid. You would most likely begin to have corruption on the screen or a BSOD if GPU temperatures were an issue. Because your PC is shutting itself down with no warning this points to an overloaded PSU. Comparing your temps to the ones I recieved while writing the review is not conducive to solving your problem. The temps I recorded were taken in an open system in an air conditioned room held at 23c ambient apx. With the recent temps it really is no surprise you are getting high GPU temps. Take a look on any enthusiast forum and everyone is moaning about high CPU and GPU temps at present.

If you do manage to sort your power issue out, I really wouldn't worry about 90c under load conditions on a GPU, running cards close together in crossfire (especially on a maximus X38) will inevitably increase temps but they can take 90c+ all day long.Quote

02-07-2009, 11:20:59

@bilko, i bet that the fps in cod waw are capped at 90fps which shows 91fps. You need to raise the maxfps to 125 or 250 for example.

in the command line, type: /com maxfps 250Quote

02-07-2009, 11:27:21

WOW, what is wrong with the hd4770 and gts250 idle power consumption in your graph! They are bothe 40watts more that the gtx285!! And consume more power than the hd4890Quote

02-07-2009, 12:03:09

Originally Posted by name='w3bbo'

CPU bottlenecking, which you will almost certainly have with 2 4890's at your current CPU & clockspeed, is the most likely culprit in why you are not seeing good FPS. You are also running with 4GB of DDR2 at a much slower speed than the test systems 6GB of DDR3 which is running 400MHz faster than yours. I don't wish to put your system down as it sounds like a sweet setup but our test setup is specifically designed not to bottleneck any product we throw at it and as such any product we review is done so under optimum conditions.
Looking at his goods (oh my), I don't think this is his problem.

Whilst using a superior cpu is always the better choice to put the emphasis on the gpu in real games, the cpu does however slant results from artificial benches, with the better exception being vantage (it doesn't seem to benefit as-much). But what can u do, u have to stick with a platform best for longevity of testing, and stay current.

I'd personally expect no bottleneck from the stuff he has due to the cpu. Imo he has something else going on.Quote

02-07-2009, 12:36:04

Hi all just passed this site for the review about vapor-X 4890

and i must say bilko is just right on somethings.

i did buy 2 xfx hd4870 in crossfire , they run 38a40 idle and 61a66 degrees fullload

i talk about the gpu diode memIO in everest because thats the one that become the hottest point

2 x - xfx hd4870 1gig crossfire with triplehead2go digital 5040x1050

gpu1a (diode displ) 34

gpu1b (diode Mem) 38 IDLE BUT FAN SET ON 50%

gpu1c (diode shader) 33

full load is highest mem temp 66

For both cards almost same temps 1 run little hotter about 2 a3 degrees


1x sapphire hd4890 vapor-x with triplehead2go digital 5040x1050

gpu 1a = 50

gpu 1b = 58 IDLE BUT FAN SET ON 50%

gpu 1c = 48

edit: i forgot this temp is with memory set back to 900mhz duh

over the full load i do not wanna talk about the 4890 vapor-x (read vapor minus xtra) the memory was over the 87 deg

i go change it for an xfx4890 tomorrow grrrr

i think they play crysis and see that the guy be vaporized in the tree and the idea was born. a bag of baked air, and i never get used to this !!! lol

so i think its a blown up bagg of air and media hype, and please not say i have worst setup or airflow or bottlenecks

i just think this test was taken with a open pc with a big fan on it

my hd2900xt in crossfire where 67 and 71 degrees full load

q6600 @ 3200 39deg

Asus P5E 40deg

4gig corsair dominator at 1160 with cooler

wd 320 sataII 33degQuote

03-07-2009, 13:41:13

Well I went out and bought a bigger PSU. A Corsair TX850W.

Now my PC doesn't crash, but guess what... SURPRISE! it's just as hot, infact it can now reach new temps. After ½ Hr of CoD5 WaW online I reached a staggering 101 Celcius (room temp 23.5 Celcius FYI w3bbo).

That is at factory settings...God alone knows what would happen if I tried to overclock these cards coz I'm not trying it!

Now I might be able to live with that, if I bought 2 GPUs that had stock cooling but when they are heralded as the new messia this side of water cooling not just by Sapphire but also you guys. I expect something better.

Even on my slow, antiquated, bottlenecked rig.

While I'm sitting writing this my idle temp is 68 Celcius and fan speeds are at 50%. Again that is factory settings. nothing I have done. That isn't what I would call super-cool.

Now don't get all deffensive again, all I'm saying is be careful how you test/review components. In the real world we have closed cabinets and might not have this years kick-ass rig...I have last years.

Thanx for the heads up about maxfps cl0ck_ed, I had forgotten all about that, I can now see a difference albeit fluctuating:worship.

Bilko out.

Updated Sys Specs:- Q9550@3.2GHz with Thermalright Ultra-120 eXtreme cooler and Noctua NF-P12 fan. ASUS Maximus Formula Republic of Gamers motherboard with X38 chipset and 4 GHZ of Kingston HyperX PC9600 RAM @ 1207MHz being cooled with a Corsair Dominator Airflow Fan.

Corsair TX850W PSU.

Vista Ultimate 64 bit OS.

Samsung SpinPoint T166 HD501LJ 500 GB.

All incased in a Aerocool Hi-Tech 7 cabinette with a 40cm fan on the side and 2 x 12cm fans, one infront and one at the back.Quote

03-07-2009, 13:59:49

No need to be this aggressive about the measured temps, I don't doubt the neutrality of w3bbo's measurements, even if they are taken in open air.

If I were you (both of you) I'd make a topic describing your case layout, preferably with a picture of it. This way we could advice on how to make it more effective, 3*120mm should be enough to keep it below 80*C at stock settings.Quote

03-07-2009, 15:24:47

I don't know why you expected temperatures to drop because you bought a bigger PSU. My advice that it was likely down to a power issue was to resolve the crashing, which you blamed on temperatures. It seems I was correct on that occasion.

I could review items in closed quarters but then folk would complain that it is still not right because it is impossible to cater for all scenrio's. Temperatures I report should only be used as a comparison to other temperatures in the graphs I show, i.e 4890 vs GTX285 etc. What runs hotter for me will likely run hotter for you. You cannot compare the temps I got to what you have as they are two totally different setups in two totally different environments I'm afraid.

Have you tried just one GPU and see what temps that reports? I have a feeling that both cards are emitting too much heat for your case to cope with, heating up the case without the exhaust fans being able to keep up.

If your cards are sitting at almost 70c idle then you either have some serious airflow problems in your case or the cards are faulty. Another review site claimed load temps which were as near as dammit the same as I got so I don't believe it's a problem with the card.

As Monkey says, you would be better making a separate post in the hardware section so people here can help you with your problem. We are afterall, only trying to help you.Quote

03-07-2009, 15:53:18

Excellent review w3bbo.

I think the moral of the story for some is to not just buy a product based on a review but to also consider one's own needs as well as considering their existing system specifications beforehand. It's also fair to say that every system even with the same specification is likely to perform a little differently and as such trying to hit the same frame rates achieved in a given review is a little unrealistic. What much of these benchmarks are good for are getting an idea of a graphics cards performance relative to another graphics card with the same testbed.Quote

03-07-2009, 20:53:14

Looking at the cooler on this vapor-x card all the heat will be exhausted into the case. Whereas the stock cooler exhausts all hot air out the back. This is probably the reason why you are getting such different results. On an open test bed the cooler will be at an advatage as its exhausted heat will be quickly transmitted away from the card.

I think testing temps of the cards on an open test bed isn't a good idea as for the majority of people it will not give a good indication of gfx card temperatures. I think you sound choose a popular case like the antec 300.Quote

03-07-2009, 21:45:31

Hi all ;-)

at first place was it not an attack on you but forme to the vapor-x

so sorry for that if feels so, you did write good review thats for sure.

i know cost lot time and sweat and tears to make all work okay

and write, make pictures ect

and i did not buy it on your review i just did buy it some hours earlier,

nothing to blame you, if i did also not

so well done job there of course , i for sure would do it not better

and i am not here complaining the review.

but my pc case is very cool but noisy

2x 25cm front and side

2x 12 cm out back and psu

fan set on memory dominator

zalman copper on processor

if i say temps 67 and 71 fullload hd2900xt crossfire,1 had an artic cooling triplefan (standard 2 heatpipes was first type was to hot to), that was 67 but the fans almost never go rotate only auto at 70 degrees,other was a new type standard with 3 heatpipes-15degrees less then 1th in begin, so i think you can see my case is very cool for that card i think.

but something is right here its an open cooler also, so not all heat go out, also a lot heat come in pc case, they claim its produced so it go out, but it always will go out ofcourse ;-)

so thats the point for now all run hot on idle because fan go to slow,under 1000 on start point, but when go to power mode its also very loud .and it also blow lot VERY HOT AIR in the pc case.

so my opinion , the cooler looks greater then the funtion it have.

maybe they look to much to try win performance but better win in less power and less heat also .

so for my opinion i think the cooler vapor-x as better cooling isnt a well done job.

also not the noise from the fan , thats also more at 40% then the xfx4870 s in crossfire on 60%

my card is 53 degrees now stock but the memory is 64 , so your temps indication are right if not look at the memory

my temps are ok then but memory go 10a15 higher look in everest it gave 3 temps 53gpu 64mem 51environment

also the clocks of memory not go back in idle mode only core go to 240.

so for me its ok so i will keep the fan noise for now better then the hot part

i did not change it because i did change already this for 2 hd4870

so i hope you understand all was not attack but just an uncontrolled impulsive reaction on the hottemps instead the review lol

have fun all nice weekendQuote

04-07-2009, 05:42:54

To avoid total offtopicness, I've taken the freedom of posting a new topic for you:

Register for the OC3D Newsletter

Subscribing to the OC3D newsletter will keep you up-to-date on the latest technology reviews, competitions and goings-on at Overclock3D. We won't share your email address with ANYONE, and we will only email you with updates on site news, reviews, and competitions and you can unsubscribe easily at any time.

Simply enter your name and email address into the box below and be sure to click on the links in the confirmation emails that will arrive in your e-mail shortly after to complete the registration.

If you run into any problems, just drop us a message on the forums.