|
Title: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: gpucoolingmethod on February 28, 2014, 09:48:32 PM Ive just found that AMD renamed its 7770 to R7 250 and its selling it as another product.
Update Ive found that all Rx are rebranded except R9 290 series.(Source http://www.tomshardware.com/reviews/radeon-r7-250x-graphics-card-review,3747.html (http://tomshardware.com/reviews/radeon-r7-250x-graphics-card-review,3747.html) . I guess thats why these arent on hashrate lists online) Am I correct to expect exact same hashrates as from the original not renamed versions? Title: Re: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: Wipeout2097 on February 28, 2014, 09:58:37 PM The R7-260 series aren't rebranded either. They have an new audio system and processor.
I've read that many cards come with faster Ram and worse power block (low cost VRMs, capacitors and so on...) They will tend to hash better, but failure rate may also be higher. For example, most 7870's have an hard time to reach 400 Kh/s while the Gigabyte and Sapphire R9-270 can touch 500 Title: Re: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: gpucoolingmethod on February 28, 2014, 10:03:01 PM The R7-260 series aren't rebranded either. They have an new audio system and processor. I've read that many cards come with faster Ram and worse power block (low cost VRMs, capacitors and so on...) They will tend to hash better, but failure rate may also be higher. For example, most 7870's have an hard time to reach 400 Kh/s while the Gigabyte and Sapphire R9-270 can touch 500 is it only because the new cards are overclocked compared to the same/similar ??? Title: Re: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: gpucoolingmethod on March 01, 2014, 07:59:32 AM The R7-260 series aren't rebranded either. They have an new audio system and processor. I've read that many cards come with faster Ram and worse power block (low cost VRMs, capacitors and so on...) They will tend to hash better, but failure rate may also be higher. For example, most 7870's have an hard time to reach 400 Kh/s while the Gigabyte and Sapphire R9-270 can touch 500 is it only because the new cards are overclocked compared to the same/similar ??? YES/NO Title: Re: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: RWatts on March 01, 2014, 12:43:32 PM I've found a lot of the memory to be of a different quality too. Generally the Rx cards perform much better for me. Keep in mind, at the factory a lot of the cards have IDENTICAL hardware, and then they test them at various clock speeds. Some work better than others, and that's often the deciding factor if a card gets branded as an OC model or not. Ultimately that's the result of imperfect manufacturing techniques. Same thing applies to processors and basically everything that uses today's ultra small fab processes.
Title: Re: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: gpucoolingmethod on March 01, 2014, 01:42:18 PM I've found a lot of the memory to be of a different quality too. Generally the Rx cards perform much better for me. Keep in mind, at the factory a lot of the cards have IDENTICAL hardware, and then they test them at various clock speeds. Some work better than others, and that's often the deciding factor if a card gets branded as an OC model or not. Ultimately that's the result of imperfect manufacturing techniques. Same thing applies to processors and basically everything that uses today's ultra small fab processes. Thats what I had in consideration i think its called binning (transistors) but I believe that they also add some safety margin just in case. Thats why you can always overclock instead of having the oveclock preapplied (notice that it wouldnt be called overclock then) AND that safety margin is the MHz that enough cards (in their opinion) work with that MHz Title: Re: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: tom_o on March 01, 2014, 05:31:54 PM Thats what I had in consideration i think its called binning (transistors) but I believe that they also add some safety margin just in case. Thats why you can always overclock instead of having the oveclock preapplied (notice that it wouldnt be called overclock then) AND that safety margin is the MHz that enough cards (in their opinion) work with that MHz The new cards do get reliably higher hashrates, but only at certain special frequencies such as 1500 or 1800 RAM on 280Xs coupled with 1080 or 1160 core respectively for 750 and 780-807 khash! Title: Re: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: gpucoolingmethod on March 01, 2014, 06:05:51 PM Thats what I had in consideration i think its called binning (transistors) but I believe that they also add some safety margin just in case. Thats why you can always overclock instead of having the oveclock preapplied (notice that it wouldnt be called overclock then) AND that safety margin is the MHz that enough cards (in their opinion) work with that MHz The new cards do get reliably higher hashrates, but only at certain special frequencies such as 1500 or 1800 RAM on 280Xs coupled with 1080 or 1160 core respectively for 750 and 780-807 khash! why certain special frequenzzies? Title: Re: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: tom_o on March 01, 2014, 06:16:50 PM I guess it's the getting the bandwidth of the RAM maximised in a way the the GPU is smoothly and constantly fed with data. Not sure why, but anything other than 1500 or 1800+ RAM is rubbish on R280X.
Title: Re: YES/NO QUESTION: are all renamed radeon hashrate exactly same as not? Post by: gpucoolingmethod on March 01, 2014, 06:26:37 PM I guess it's the getting the bandwidth of the RAM maximised in a way the the GPU is smoothly and constantly fed with data. Not sure why, but anything other than 1500 or 1800+ RAM is rubbish on R280X. I think thats not true because either the memory or core waits for the other unless they are timed perfectly.But whatever waits doesnt make a difference if the clock rates are equivalent to what you are comparing.THE REAL REASON IS: bad memory latencies/timings. They are coded as different values for different speeds and they very often are suboptimal. how can one find out better than that? |