MSI R5870 Lightning Video Card Review @ [H]

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,724
MSI R5870 Lightning Video Card Review - Today we have MSI's latest Lightning series video card, the R5870 Lightning on our test bench. Is it an explosive success, a shocking failure, or none of the above? We checked it out with the help of six of today's hottest games, and we're here to tell you what we found. The results just might surprise you!
 
Thanks for the review Kyle! What a disappointment :(

*hugs Sapphire Vapor-X 5870*
 
Thanks for the review. Hard to believe that OC considering the PCB and high-end parts they're using. I can imagine how many people would be mislead, if not for this review.
 
I've never said this about any review here on the [H] but I question your results here with the overclocking. Please note that I have no stake in this video card or MSI in general, but your results fly in the face of what I have seen on just about every other 5870:

Using this software, our maximum stable overclock was a paltry 915MHz on the GPU and 1225MHz (or 4900MHz effective) on the memory. To accomplish this, we had to increase GPU core voltage to 1159mV, or 1.159 Volts. Increasing the voltage further did not allow us to achieve a higher overclock.
Click to expand...
You are saying that you could not achieve a more stable overclock even with the extra .19v headroom afforded in Afterburner? My 5870s run at 1.16v load under their normal BIOS. The ASUS EAH5870 V2 comes at 1.21v. You should be able to turn the card all the way up to 1.35v.

With reasonable cooling you should definitely be able to hit much higher than 915MHz. How high did you test, and did you scale the fan profiles accordingly (or did you touch the fan scaling at all or just lock it at a higher speed)? That's one of the best functions of the Afterburner tool. Or do you test the overclocking capabilities with the default fan scaling of the card in question?

I just find it hard to believe that you could not eek more than 915MHz out of the card, as even my reference boards do 930-950MHz, and that you did not see gains with increasing the voltage.
 
Rizen said:
I've never said this about any review here on the [H] but I question your results here with the overclocking. Please note that I have no stake in this video card or MSI in general, but your results fly in the face of what I have seen on just about every other 5870:

You are saying that you could not achieve a more stable overclock even with the extra .19v headroom afforded in Afterburner? My 5870s run at 1.16v load under their normal BIOS. The ASUS EAH5870 V2 comes at 1.21v. You should be able to turn the card all the way up to 1.35v.

With reasonable cooling you should definitely be able to hit much higher than 915MHz. How high did you test, and did you scale the fan profiles accordingly (or did you touch the fan scaling at all or just lock it at a higher speed)? That's one of the best functions of the Afterburner tool. Or do you test the overclocking capabilities with the default fan scaling of the card in question?

I just find it hard to believe that you could not eek more than 915MHz out of the card, as even my reference boards do 930-950MHz, and that you did not see gains with increasing the voltage.
Click to expand...

I tried multiple fan configurations. I made it scale neatly with temperature, then set it to100% at all times, and so on. Nothing helped.

Increasing the voltage did not help me increase clock speed on this particular unit. I don't know how more clearly I can state this.

Overclocking on this card was definitely a non-starter. I even tried different versions of Afterburner and even the Catalyst driver. I cannot say that every R5870 Lightning will have this problem, but this specific card did.
 
Mark_Warner said:
I tried multiple fan configurations. I made it scale neatly with temperature, then set it to100% at all times, and so on. Nothing helped.

Increasing the voltage did not help me increase clock speed on this particular unit. I don't know how more clearly I can state this.

Overclocking on this card was definitely a non-starter. I even tried different versions of Afterburner and even the Catalyst driver. I cannot say that ever R5870 Lightning will have this problem, but this specific card did.
Click to expand...
Well damn. You got a dud of a card then :( Did MSI have any comment on those results?
 
I'm finding these "special edition" cards special alright. About 90 to 100 dollars more special for nothing really special.

I'm ready for the fall refresh so a stock 5870 will come down to a reasonable price. These overclocked cards just don't make me feel "special". :rolleyes:
 
dceremuga said:
I'm finding these "special edition" cards special alright. About 90 to 100 dollars more special for nothing really special.

I'm ready for the fall refresh so a stock 5870 will come down to a reasonable price. These overclocked cards just don't make me feel "special". :rolleyes:
Click to expand...

I suspect that MSI's "Military Class" hardware will probably physically last longer than most reference designs, but there are still problems. Even if the capacitors and chokes are rated for a bazillion times more hours of life, you still have to deal with the HDMI, DVI, DP, and other connectors, which are subject to the same wear and tear as every other connector. And my personal experience has been that connectors are some of the weakest points.
 
nice review.. but my question is why are companies still coming out with these 1GB cards? why not just release 2GB versions with the same 4 display ports.. gain slightly better performance in single display and bigger gain in multi-display.. its not like everyone needs a 6 DP card.. i just dont understand why they havent scrapped the reference 1GB version for a non reference 2GB version.. i just see it as a waste of time and money for these companies to keep releasing these things.. because i sure as heck wouldnt mind paying a 90 dollar premium for a 2GB version with 2 dvi + DP and hdmi instead of 6 DP's..
 
Rizen said:
I've never said this about any review here on the [H] but I question your results here with the overclocking. Please note that I have no stake in this video card or MSI in general, but your results fly in the face of what I have seen on just about every other 5870:

You are saying that you could not achieve a more stable overclock even with the extra .19v headroom afforded in Afterburner? My 5870s run at 1.16v load under their normal BIOS. The ASUS EAH5870 V2 comes at 1.21v. You should be able to turn the card all the way up to 1.35v.

With reasonable cooling you should definitely be able to hit much higher than 915MHz. How high did you test, and did you scale the fan profiles accordingly (or did you touch the fan scaling at all or just lock it at a higher speed)? That's one of the best functions of the Afterburner tool. Or do you test the overclocking capabilities with the default fan scaling of the card in question?

I just find it hard to believe that you could not eek more than 915MHz out of the card, as even my reference boards do 930-950MHz, and that you did not see gains with increasing the voltage.
Click to expand...

To further Mark's comments, we also took the card apart, cleaned it up real well, applied our own thermal paste in the appropriate amount and put it all back together well and tried overclocking again. So we did overclock testing out-of-box, then we did overclock testing after having taken it apart and cleaned it up, and used a newer version of Afterburner as well as Cat 9.6, and the results were still the same. We tried mighty hard to make it overclock as much as possible, it just wouldn't. Our testing was quite thorough in regards to overclocking, we understand it is a key feature of this video card.

It should also be noted, looking back at our GeForce 275 GTX Lightning Video Card Review that we also did not get the overclocks we had expected - http://www.hardocp.com/article/2009/09/14/msi_n275gtx_lightning/9
 
Last edited:
The HIS H587FN1GD Radeon HD 5870 1GB is only $350.99 at newegg with free shipping if you use promo code VGA061710 for 10% off.
 
Brent_Justice said:
To further Mark's comments, we also took the card apart, cleaned it up real well, applied our own thermal paste in the appropriate amount and put it all back together well and tried overclocking again. So we did overclock testing out-of-box, then we did overclock testing after having taken it apart and cleaned it up, and used a newer version of Afterburner as well as Cat 9.6, and the results were still the same. We tried mighty hard to make it overclock as much as possible, it just wouldn't. Our testing was quite thorough in regards to overclocking, we understand it is a key feature of this video card.

It should also be noted, looking back at our GeForce 275 GTX Lightning Video Card Review that we also did not get the overclocks we had expected - http://www.hardocp.com/article/2009/09/14/msi_n275gtx_lightning/9
Click to expand...
Okay :) I don't doubt your technical capabilities, I was just surprised that the results were that bad. Even a reference card without the enhanced cooling or components of the MSI board typically will hit a higher clock speed.

Do you think you just got a bad board, "luck of the draw" so to speak? Or do you think this might be something related to MSI's design? Just curious.
 
I was reading some post in other asian forums, it appears there is a second revision to this particular model.

The second revision have way better cooling parts that is redesign, which if you contact MSI, they will give you the newly redesign fan for free.

Not quite sure about how second revision model overclock, but seems like a lot of people were having problems with first revision.
 
Rizen said:
Okay :) I don't doubt your technical capabilities, I was just surprised that the results were that bad. Even a reference card without the enhanced cooling or components of the MSI board typically will hit a higher clock speed.

Do you think you just got a bad board, "luck of the draw" so to speak? Or do you think this might be something related to MSI's design? Just curious.
Click to expand...

Have no clue, our card though is a fully retail packaged card, just as the consumer would buy on the shelf or online. I wish we could test 100 cards for overclocking and find an average, but that just isn't possible. Still, if this card were on the shelf, and you picked it up, this is what you'd get on this card for OC. I think MSI has designed this card to last, the parts are all high quality so that it is robust and will last longer than other cards using more standard parts, it should also be able to take more of a beating on the components when you do do things like push voltages and clock frequencies. Think of it more perhaps as the "Panasonic Toughbook" of video cards? Though I wouldn't recommend dropping it on the ground ;)
 
Thanks for the review, nice to see the first [h] nvidia vs ati post 257 driver release, w00p.

Non ref 58X0s seem to be pretty smelly overclockers in general, that asus ROG 5870 is the exception that proves the rule :p.
 
thanks for the review guys - got a couple questions though.

I was curious if you used a multimeter to see if afterburner was actually changing the voltages. some of the earlier bios had issues.
the stock voltage on my lightning is 1164 in afterburner and you achieved your overclock results at 1159...? my card hits 960/1250 with the default 1164.

seeing so many mixed results from reviews and other users alike, i do believe your OC numbers are right. this card seems extremely hit or miss.

also, thanks for posting the lightning tweaker thing. i always wondered why the lights never changed on the back of my card. i had no idea I had to run that software for it to work :D
 
Brent_Justice said:
To further Mark's comments, we also took the card apart, cleaned it up real well, applied our own thermal paste in the appropriate amount and put it all back together well and tried overclocking again. So we did overclock testing out-of-box, then we did overclock testing after having taken it apart and cleaned it up, and used a newer version of Afterburner as well as Cat 9.6, and the results were still the same. We tried mighty hard to make it overclock as much as possible, it just wouldn't. Our testing was quite thorough in regards to overclocking, we understand it is a key feature of this video card.
Click to expand...
Well, thanks to you and Mark for going through all the trouble. No one can say you didn't try.

It's sooo pretty, though...
 
I very much appreciate this critical review. I was about to order a system to be built using the MSI Big Bang XPower motherboard, which I believe also uses some of the same types of "highly rated military components" used in this card. However, with only 3 reviews from NewEgg.com :

http://www.newegg.com/Product/Product.aspx?Item=N82E16813130289&Tpk=msi big bang xpower

and after reading this review, I think I will hold off for a few weeks until more user-reviews have been posted.

I saw your sneak peak on this motherboard a couple of days ago and I'm assuming that you are going to be publishing a full review sometime in the future. If this is true, can you give us an idea of when that review might be published?

This is definitely a very disappointing review for this video card, but the upside is that MSI should be quite motivated to either fix the problem or improve the perforance of the card. If they were to send you a revised version of the board, would you be open to retesting it and publishing the new results? I understand that you probably have too much to do already with all the products out there, but I too would like to know if this particular card is a "dud" as a couple of the other users here suggested, or whether there is a more serious general flaw in the board that is possibly linked to the underlying components and whether or not that might affect some of their other products like the MSI Big Bang XPower motherboard.

Any chance you could send the video card back to them and ask them to test it for defects? If they come back and say that this is a good card and fairly represents their product line, then your readers will know that if they buy this product, they will most likely receive the same kind of performance. If instead they come back and say that the card is defective somehow, then this review does not mean a whole lot and becomes irrelevant because it is not representative of what most others will be obtaining when they install this card in their system.

Lastly, is there any chance that when you do a motherboard review you could try out different memory sticks from different manufacturers and then do your testing with the memory that yields the best performance or best performance at a reasonable price? I have read comments of others who say that installing faster rated memory in a system is a waste of money and has very little effect. But, on other sites that rate memory, I have seen graphs that show better performance and better fps in games. I don't know who is right...
 
Kyle, Mark, or Brent, I noticed that you used the 257.15 drivers for the GTX480 (I am guessing you didn't get the 257.21 in time for this review). Do you think that the use of the 257.21 would put any more distance between the 480 and 5870? (in your review here that is.) Right now I can get a power color 5870 oc (950mhz) for 390 shipped from newegg. I was hopping to see the GTX480 pull out a little. not to mention this kills the MSI value wise.
 
Valset said:
Kyle, Mark, or Brent, I noticed that you used the 257.15 drivers for the GTX480 (I am guessing you didn't get the 257.21 in time for this review). Do you think that the use of the 257.21 would put any more distance between the 480 and 5870? (in your review here that is.) Right now I can get a power color 5870 oc (950mhz) for 390 shipped from newegg. I was hopping to see the GTX480 pull out a little. not to mention this kills the MSI value wise.
Click to expand...

Probably not, the gains reported by NV are the same. A few bug fixes and WHQL is the difference.
 
Heh, mine OC'd better then yours did :)

I'm at the point where I think all the custom cards are a crap shoot. Some of them are golden, others no better then teh standard.
 
Not all chips are made equal. Obviously some 5870s will barely break stock while many others have been seen to break 1000MHz.

I'm just disappointed that for a premium card where so much was spent on other parts of the design and components that MSI didn't take the simple step of binning the core that went into it.
 
Spare-Flair said:
Not all chips are made equal. Obviously some 5870s will barely break stock while many others have been seen to break 1000MHz.

I'm just disappointed that for a premium card where so much was spent on other parts of the design and components that MSI didn't take the simple step of binning the core that went into it.
Click to expand...

Since posting yesterday and not getting any replies, I decided to do a little more research and read some reviews on other sites. I won't post any links because I don't want to piss off anyone here. But, a review in the UK published Jun 15 2010 claimed that he was able to manually overclock the core clock frequency of this card to 1010 mhz from the stock 900mhz. For the memory he claimed he was able to get 5860 mhz from the stock 4800 mhz. This resulted in an increase of about 20% in fps for 2 games - Far Cry 2 and Alien vs Predator. His system is liquid cooled though, and that might make a difference. His conclusion is that this card is the best 5870 clone on the market today. Another reviewer with an air cooled system published a review on Apr 12, 2010 and claimed overclocking to 1034 mhz on the core and 5404 mhz on the memory. A 3rd review published on Mar 17, 2010 with an air cooled system claimed overclocking speeds of 1000 mhz on the core and 5000 mhz on the memory.

Based upon the other reviews I have read, it just seems to me that the most likely cause of the problems they had with overclocking is that they got a bad card, memory, motherboard, etc. It happens. Ideally, it would be nice if they could obtain 3 cards from the manufacturer and then perform a quick test on each. Throw out the top performing card and the weakest card and then do the in depth testing with the middle card.
 
RobertGBryan said:
Since posting yesterday and not getting any replies, I decided to do a little more research and read some reviews on other sites. I won't post any links because I don't want to piss off anyone here. But, a review in the UK published Jun 15 2010 claimed that he was able to manually overclock the core clock frequency of this card to 1010 mhz from the stock 900mhz. For the memory he claimed he was able to get 5860 mhz from the stock 4800 mhz. This resulted in an increase of about 20% in fps for 2 games - Far Cry 2 and Alien vs Predator. His system is liquid cooled though, and that might make a difference. His conclusion is that this card is the best 5870 clone on the market today. Another reviewer with an air cooled system published a review on Apr 12, 2010 and claimed overclocking to 1034 mhz on the core and 5404 mhz on the memory. A 3rd review published on Mar 17, 2010 with an air cooled system claimed overclocking speeds of 1000 mhz on the core and 5000 mhz on the memory.

Based upon the other reviews I have read, it just seems to me that the most likely cause of the problems they had with overclocking is that they got a bad card, memory, motherboard, etc. It happens. Ideally, it would be nice if they could obtain 3 cards from the manufacturer and then perform a quick test on each. Throw out the top performing card and the weakest card and then do the in depth testing with the middle card.
Click to expand...

The card has been claim to be cherry pick for reviewers. That is why they can hit such good OC while consumers can't.

I am begin to wonder if [H] got this card from retailers other than MSI themselves..
 
RobertGBryan said:
Since posting yesterday and not getting any replies, I decided to do a little more research and read some reviews on other sites. I won't post any links because I don't want to piss off anyone here. But, a review in the UK published Jun 15 2010 claimed that he was able to manually overclock the core clock frequency of this card to 1010 mhz from the stock 900mhz. For the memory he claimed he was able to get 5860 mhz from the stock 4800 mhz. This resulted in an increase of about 20% in fps for 2 games - Far Cry 2 and Alien vs Predator. His system is liquid cooled though, and that might make a difference. His conclusion is that this card is the best 5870 clone on the market today. Another reviewer with an air cooled system published a review on Apr 12, 2010 and claimed overclocking to 1034 mhz on the core and 5404 mhz on the memory. A 3rd review published on Mar 17, 2010 with an air cooled system claimed overclocking speeds of 1000 mhz on the core and 5000 mhz on the memory.

Based upon the other reviews I have read, it just seems to me that the most likely cause of the problems they had with overclocking is that they got a bad card, memory, motherboard, etc. It happens. Ideally, it would be nice if they could obtain 3 cards from the manufacturer and then perform a quick test on each. Throw out the top performing card and the weakest card and then do the in depth testing with the middle card.
Click to expand...
A lot of the [H] reviews are done on hardware they purchase outright themselves, IIRC. Buying 3x $400 video cards for every review would probably be cost prohibitive.
 
no pb to clock mine! all on air
my lightning only needs 1.174v to hit 1000/1350
and 1.299v to hit 1100/1350:

the only thing needed: spacers for perfect gpu contact with the heatsink
 
1100MHz on air at 1.29v is phenomenal. You got a one in a million chip there.
 
Thanks for the review, [H].

Less than stellar performance indeed. You all had to have unfortunately gotten a dud. Many of the other reviews are pumping the card up higher without too much effort.

Despite the overclocking issue, the military grade components are a nice plus for those who are looking for more long-term usage than anything else.
 
Rizen said:
1100MHz on air at 1.29v is phenomenal. You got a one in a million chip there.
Click to expand...

He probably bought the new Lightning II which fixes the voltage and overclock problem. MSI quietly rolled it out since there was a problem with the original Lightning. All of the reviews for the Lightning II are giving it praise for its overclocking potential.

Kyle if you read this I have read on another forum that if you contact MSI they will swap out your old card for the Lightning II. I don't think it would be very fair to give a bad review on a product because you got a bad batch.

Newegg link for the new version.
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127509
 
Snyper64 said:
He probably bought the new Lightning II which fixes the voltage and overclock problem. MSI quietly rolled it out since there was a problem with the original Lightning. All of the reviews for the Lightning II are giving it praise for its overclocking potential.

Kyle if you read this I have read on another forum that if you contact MSI they will swap out your old card for the Lightning II. I don't think it would be very fair to give a bad review on a product because you got a bad batch.

Newegg link for the new version.
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127509
Click to expand...

I don't know that the people that bought the first one would agree. I doubt the company is going to swap them out.
 
my card is not a new revision of the lightning! it's the first released ;)
to hit 1100mhz i only add spacers in order to improve gpu contact!
Techpowerup reported this issue...
 
You must log in or register to reply here.
Back
Top