Sapphire HD 7950 Vapor-X 3072 MB

[W1zzard]

To read this review go to: http://www.techpowerup.com/reviews/S..._7950_Vapor-X/

[Dj-ElectriC]

An HD7950 that goes 1135Mhz core (41% over stock 7950 clock) at stock voltage.
People should freaking riot in the streets. As much as this is amazing, i am not surprised, Sapphire after all.

Please w1zzard, share some over-voltage OC results.

[cadaveca]

This card is basically using refinded 7970 PCB, including 8-pin + 6-pin, so OC is not surprising, to me.

Which sucks, I'd like ot try a 950 MHz boost BIOS, but VRM section is different than my 7950s!

[Casecutter]

With a great construction and components it makes it an undeniable value at that MSRP. Though consider Sapphire has been aggressive right out of the gate on some of these Vapor-X with rebates. Definitely a top-shelf consideration that’s easy-going on the wallet.

[happita]

Still slower and draws more power than a 670 at stock The only good you get out of these specialized cards is the OC potential which then exceeds MOST of the other cards in performance. And the cooling solution is GREAT for low noise environments. Now this should have been the kind of performance we should have gotten when the 7950 cards were first released.
But then, what would manufacturer's do to milk the series for all its worth? Oh, that's right, initially release cards at a subprime clock speed then re-release the cards at higher frequencies and charge another premium over reference. I hate that ideology, but I guess thats business
Anyway, nice review W1z, always a pleasure looking at the numbers you come up with

[eidairaman1]

This would be the card I would get Honestly less if MSI had a Hawk/Lightening Edition

[HumanSmoke]

Originally Posted by Casecutter

Definitely a top-shelf consideration that’s easy-going on the wallet.

Nice to see some people getting the benefit of overlapping AIB distribution. Around here, the HD 7950 ( non-boost version) retails about 20% higher than a GTX 660 Ti, and around the same price as the GTX 670. If you're willing to drop to a reference 670, you can pick up a Palit/Galaxy/Leadtek for around 10% less than a reduced-BoM Sapphire 7950 Flex

[Casecutter]

Originally Posted by HumanSmoke

HD 7950 (non-boost version) retails about 20% higher than a GTX 660 Ti,

Wow, that stinks... here state side a Sapphire 100352OCSR is 15% less than the cheapest 670 on a smurf board. While an XFX Double D FX-795A-TDJC is $290 -AR$30. The lowest GTX670 I've seem is for an EVGA 02G-P4-2670-KR $375 -$20 off coupon (click "Clip this coupon")= $355 with free shipping at Amazon.

Heck this Sapphire Vapor-X 100352VXSR (950MHz Boost State) is now $330 8.50 to ship.

[Jacob XP]

Great review thanks.

Power consumption in idle is EXTREMELY high. It's a little over 30 watt... A reference 7950 is 11 watt.

Even in Northern Europe where electricity prices due to taxes is very high it's not really that expensive to power the 7950. I can easily afford the few $ extra that it costs

But what where they thinking when they designed the card. Some engineers have to hide in shame if you ask me.

Would it be possible to use some tricks to lower the consumption a bit. Could you lower the voltage and still get a stable card?

[tt_martin]

So, I've checked performance of GTX660ti on the launch review and this - it's the same. @W1zz why is that 660ti in overall performance is slower (by 1%) than 7950 then and now it's 1% faster (1920x1200p res)?

[DarkOCean]

the oc gain its just amazing as the graphics card and the review .

[Mathragh]

Whoah! Those clocks are insane for that voltage! My reference 7950 doesnt even come close to those clocks at the stock(1.07) voltage.

If only the power consumption was lower to the same degree.

[tt_martin]

Originally Posted by DarkOCean

the oc gain its just amazing as the graphics card and the review .

Yep and people still prefer GTX 660Ti over HD 7950

[DrunkenMafia]

I just bought a powercolor 7950 and so far it runs fine @ 1020mhz with no voltage increase, just played some bf3 with no hassle and temps never above 60c with stock cooling.. waiting for a block for the card so I can up the voltage a little.

Awesome card though, kicks crap out of my 5870 crossfire setup and cost the same as selling my 5870's on ebay.

[W1zzard]

Originally Posted by tt_martin

So, I've checked performance of GTX660ti on the launch review and this - it's the same. @W1zz why is that 660ti in overall performance is slower (by 1%) than 7950 then and now it's 1% faster (1920x1200p res)?

could be because wow was dropped from our benches. they released a new patch which invalidates all previous results

[tt_martin]

Originally Posted by W1zzard

could be because wow was dropped from our benches. they released a new patch which invalidates all previous results

Still, 660ti was faster in wow than 7950 and overall it's result should be worse, not better.

[DarkOCean]

Originally Posted by tt_martin

Yep and people still prefer GTX 660Ti over HD 7950

That's an example of good marketing from nvidia and to make things worse just look at the metro 2033 results.

[tt_martin]

One thing, this 7950 VaporX Boost BIOS works really bad.
7950@850Mhz is 4% faster than stock 7950, diff in clocks - 6.25%
but with Boost we gain only 3% over 850Mhz, diff in clocks - 11.76%

We should get better results with manual overclocking

[Mathragh]

Originally Posted by tt_martin

One thing, this 7950 VaporX Boost BIOS works really bad.
7950@850Mhz is 4% faster than stock 7950, diff in clocks - 6.25%
but with Boost we gain only 3% over 850Mhz, diff in clocks - 11.76%

We should get better results with manual overclocking

The lack of a memory overclock in the boost bios could probably explain that

[INSTG8R]

OH how I wished I waited just a LITTLE longer for the Vapor-X

Loved my 4870 and 5870 Vapor-X's These ones are looking just as good(As I would expect) I mean I waited until June to finally bite the bullet. 6 months, I figured they weren't coming and were just going with the "Dual X" coolers this round. Ah well lesson learned. I get 1100/1500 just using Overdrive so I shouldn't complain.

[billcat479]

I hate that ideology, but I guess that's business
Anyway, nice review W1z, always a pleasure looking at the numbers you come up with [/QUOTE]

I'm sorry because I don't like flame but here it is for what it's worth. And I really am sorry to type this crap.
The clocks that certain cards come with (so I read and so I hope) are special tested GPU's that can run with stability at higher clock ratings and so they go into the special package lineup for each company that does it.
If they did it hit and miss or every chip could do it then they would all have higher clock settings starting with the reference cards because they know they can do it without a load of failed cards.
As the people or companies that buy the GPU's in large numbers they run their own tests to find GPU's that they feel can run at higher clocks with no errors or failures. They also try to make them last even longer by putting a better heatsink/fan so the card will work for at least 2 years or so.
But they would not use untested GPUs for a lot of obvious reason and these reasons end up costing them and us more for these types of cards because it makes the end user very mad at any company that sells a card that has a high failure rate, they want their customers to be happy with their products and in this segment of the industry it is really important to have a trusted name.
No company in their right mind would do this for a lot of reasons, mostly credibility because without that they don't sell products.
To search though a batch of GPU's that they buy takes time which is money so when they do put out the higher clocked ones it's reasonable to charge higher prices for their effort in finding GPU's that will be stable 100% of the time versus random ones that everyone gets and might get away with overclocking as high but have to either raise the voltage and put better cooling on or run the existing setup fan speeds at a high level with corresponding noise that no one likes and still they may see artifacts or glitches and shorten the life of their cards in the process.
And please, don't come in being a Nvidia ass kisser ok? The people that read this are smart enough to make up their own minds about which brand they want to run.
And the Nvidia cards are really not much if any faster across the line because with the new driver software the 7900 series beats it in a few game and losses in others and they are both pretty damm close to each other performance wise. They are both good cards and no one would be put out buying either one so do more reading yourself before putting you keyboard like foot in your mouth and don't come off sounding like (ugh, this I hate) a Nvidia fan only even if the facts don't bear it out.
It's cool to prefer brands and support the company so it stays in business but that should always be left at home and not thrown all over the web. If you actually think one is better then give links and proof to back it up. I've read enough to know they are pretty toe to toe performance wise and cost is really the factor in buying one over the other.
Again, very sorry to sound cruel, I don't like doing this but I really hate fan posts because it just puts mud up there to confuse people that are after credible information on what to buy.
And the second problem is using a test rig that is not what people use so the data given doesn't help anyone one bit.
If someone has a cpu clocked at 3.2GHZ and thinks they will get the same performance and goes out and buys it only to find they should have gotten a better card because they see the numbers and think they can come close or see lag spikes instead then they just learned they can't trust web site benchmarks because they really are useless. I've only seen one site that did a real comp. test of cpu/gpu performance and the results were quite a eye opener. You may get the FPS that is shown on tests like this one but it's a very very fluctuating graph {not shown here }and gives no data on actual playability.
I've seen it enough to know this is a very important area that no one (except this site called the Tech Report, yeah the name is close to this site but they are different sites) really gave a huge going over with different cpu's at different speeds with various popular video cards and the data given showed that the basic FPS on most reviews is absolutely useless because they never STAY at that speed and most have mill-second spikes that you notice at times like the game gets a little jerky even though you can test it and get 80FPS but it's still a jerky game in areas that the tested data says it will not slip down below 20FPS on a regular basis but with some computers this happens quite a bit more than is comfortable to see.

For the most part a lot of the tests are just hit enticement for the websites income and they spend as little time as they can on these tests to bring in people to get the hit count up and that is the sad reality of the information super highway we call the web or internet where 1% is intelligent and .05% is useful while the rest is filler and disinformation with money being the driving factor in the end and that is down right sad. You really do have to be smart in the first place to be able to filter the useless from the useful.
For what it's worth to this site, it IS a good site and it may not have the time to do a good enough review and just give the basics but everyone is doing this and the readers really are not getting useful information to take to the store and pick out the right video card and this really makes me mad because video cards are not cheap and when people may have to save up money to buy one and then finds out it's not what they needed and end up blowing a lot of money.
I got zapped this way and ended up going crossfire because I didn't want to go buy the next faster card and waste the one I did buy so I got another one and ran it that way for a while and I hate going the SLI-Crossfire route because there always seems to be issues with some software that they don't work all that well and I like to keep things simple. I was mad which is why I added my dislike of the benchmark testing done here and just about everywhere else. And I'm not the only one that I know that has fallen into this trap either and I don't know that many people, that in it's self means a lot.

[eidairaman1]

Billcat- that was too long of a post to read, but on a point- performance numbers will always have variables whether they are .0001% or higher. What is posted is the average a user would get from such a system. I understand most don't read the entire review but it would be recommended to because system spec details, that is where the variables come in I recall the test bed was a core i7 920 but it switched to a core i7 2600 several months back.

[billcat479]

Originally Posted by eidairaman1

This would be the card I would get Honestly less if MSI had a Hawk/Lightening Edition

Hi, I got MSI 7950 TwinFrozr card and I'm very happy with it. It comes with a 880ghz GPU and right now I have it set at 950ghz with no voltage raise but I did play with the fan setting graph so it cools when it needs it (love their overclocking software, that fan custom adjusting setup is great).
But I would recommend this MSI video card to anyone looking for one that is priced ok compared to the 7970 and it can be overclocked to out perform the stock 7970 model which to me means a lot. I have taken it up to 1,000ghz and it worked fine but I don't like pressing my luck or blowing out a new card and as of yet I haven't needed it to be any faster with all the games I have.
But It's nice to know there is more room if I need it in the future.

MSI put a real good cooling package on it, it sticks around 50-55c playing Skyrim and Battlefield3 I think and Crysis's games so it can handle the best games with high graphic's settings. If you can't afford or don't have a lot of spare money it's a very good video card and you could get a few years use out of it and if worse comes to worse there is still crossfire. Me I hate messing with that setup but to each their own ect..ect... Good luck, hope you find what your looking for and it serves you well.
P.S. This MSI also came with free games though I still haven't tried to get them but I should get around to it one of these days...

[Jacob XP]

I have read some other reviews about the power consumption in idle. It doesn't look that bad

Kitguru have only tested the card in boost state and they report a graphics card ( not the whole system ) consumption of 199 watt in gaming. The reference AMD7950 is measured at 187 watt. In zero core state they both consume 4 watt. Sadly no idle reporting

Hexus has also only tested in boost state and they report a system power consumption in idle of 59 watt vs 58 watt of the reference 7950

Tweaktown is reporting a system consumption in idle at 196 Watt. It's in the same range of other cards including a Geforce 660 which I thought would have a easy win in this discipline.

I'm not saying that TechPowerUp has anything wrong while testing the power consumption card in idle. You have your test setup and other sites have a different setup. And what exactly is idle when they test the card ? I took a chance a bought the card actually. I found a big company in Denmark that wanted to sell it a a very good price when the get their first shipment. I should have the card by Friday / Saturday. I might even do my own little power test if anybody is interested. I only have some cheap power consumption tester though.

[W1zzard]

Originally Posted by Jacob XP

I have read some other reviews about the power consumption in idle. It doesn't look that bad

Kitguru have only tested the card in boost state and they report a graphics card ( not the whole system ) consumption of 199 watt in gaming. The reference AMD7950 is measured at 187 watt. In zero core state they both consume 4 watt. Sadly no idle reporting

Hexus has also only tested in boost state and they report a system power consumption in idle of 59 watt vs 58 watt of the reference 7950

Tweaktown is reporting a system consumption in idle at 196 Watt. It's in the same range of other cards including a Geforce 660 which I thought would have a easy win in this discipline.

I'm not saying that TechPowerUp has anything wrong while testing the power consumption card in idle. You have your test setup and other sites have a different setup. And what exactly is idle when they test the card ? I took a chance a bought the card actually. I found a big company in Denmark that wanted to sell it a a very good price when the get their first shipment. I should have the card by Friday / Saturday. I might even do my own little power test if anybody is interested. I only have some cheap power consumption tester though.

So you are saying, we are the only ones who tested it, and you like the results of people better who use invalid test methods. testing whole system's power consumption with a $10 power meter is not really reliable

edit: i'm crawling out of bed to restest idle just for you

edit2: the higher idle power consumption is because the card runs 0.945 V idle instead of 0.85 V like other HD 7950 cards, as mentioned in the power consumption page. just verified the voltage again using gpuz and multimeter measurement

edit3: retesting done



please note how the idle power consumption drifts up slightly after power up while the card reaches temperature equilibrium. the higher levels near the end are when i connected a second monitor for multi-monitor power consumption. it drifts up here, too because higher power = more heat = warmer card over time = higher power consumption

nevertheless, i think you'll be very happy with the card. it's a good product