Nvidia's driver automatically prioritises the faster RAM, only encroaching into the slower partition if it absolutely has to. The GTX 970 has 3.5GB in one partition, operating at 196GB/s, with 512MB of much slower 28GB/s RAM in a second partition. The GTX 980 has 4GB of GDDR5 in one physical block, rated for 224GB/s. However, the major issue concerns the onboard memory. The GTX 970 had 56 ROPs, not 64, while its L2 cache was 1.75MB, not 2MB. Anandtech's article goes into more depth on this, but other changes came to light some months later. Otherwise, it's the same tech - or so we were told. It's based on the same architecture, uses the same GM204 silicon, but sees CUDA cores reduced from 2048 to 1664, while clock speeds are pared back from a maximum 1216MHz on the GTX 980 to 1178MHz on the cheaper card. Nvidia's reviewers' guide painted a picture of the GTX 970 as a modestly cut-down version of its more expensive sibling, the GTX 980. Let's quickly recap what went wrong here. Nvidia released inaccurate specs to the press, resulting in a class action lawsuit for "deceptive conduct". But recent events have taken the sheen off this remarkable product. It beat the R9 290 and R9 290X, forcing AMD to instigate major price-cuts, while still providing the lion's share of the performance of the much more expensive GTX 980. Indeed, we called it "the GPU that nukes almost the entire high-end graphics card market from orbit". Tom's Hardware hasn't reviewed any DRAM with IC's over DDR4-2666, so that only leaves overclocking.When the GTX 970 launched last year, the tech press - Digital Foundry included - were unanimous in their praise for Nvidia's new hardware. Of course we haven't tested every kit on the market, so.prove me wrong? In all of those cases, you will always find that the top JEDEC (SPD table) configuration is also the set for which the ICs are rated. Or you can Google another article where someone either got a decent photo OR used a program, and then have the owner of that program threaten to sue you for not licensing the program even though you don't have a copy. This works pretty well for review sites but they might not respond to individuals. More than half the time, more than half the print comes off with the adhesive, and then you can read it with a magnifying glass and a flashlight but can't get a legible photo. TJ Hooker said:Where do you see the original rated speed of the DRAM chips? Are you just going off the base JEDEC profile that's included?Nah, you just peel the heat spreaders off and then when you find torn adhesive all over the place you get embarrassed and don't take photos. But since Intel was the actual IC manufacturer, you could instead say "improperly validated", and that's closer to the argument you're making. Intel had used a raised vCore to get that CPU to that frequency, and the consensus in the enthusiast community was "overclocked too far". Adata overclocked it.įrom another angle: If the chips on these modules weren't being overclocked, XMP wouldn't be necessary.Īs for CPUs, do you remember when Intel tried to release a Pentium III 1.13GHz using the Coppermine core? And remember the math error? Uncovering that put Tom's Hardware back on the map. Moving up to the present, the memory referenced in this article was originally Samsung DDR4-2666, sold by Adata as DDR4-3200. Regarding that old memory, Samsung manufactured it. Let me put it this way: Remember when OCZ sold bone-stock Samsung PC-800 RDRAM as PC-1066? And remember that it overheated on most systems? And remember what we said about it back then? You might not have been around back then, but the consensus was "overclocked too far". The RAM is indeed DDR4-2666 regardless of how the company that puts it on a DIMM advertises it. It doesn't matter who pushes it to a higher level, it's still a higher-than-rated clock. That's why moist can be "overclocked" to more than that.The manufacturer of the RAM ICs says it's DDR4-2666. JEDEC is also similar but may include even more detailed settings.Ĭonversely, if CPU can run from a set minimum frequency to it's maximum frequency on all cores is also just a set of steps in frequency and not real overclock but just normal operation within given (projected) parameters set by manufacturer to their least successful sample. Some RAM has several such XMP profiles that are just set of instructions for BIOS to set memory to certain frequency including other settings like Cas and voltage for instance for it to work at all or at it's best performance. That is not a traditional overclock that meant going over manufacturer's recommended frequencies. CountMike said:Well, I wouldn't consider XMP as a real overclock but just as a maximum clock (frequency) manufacturer considers as full speed for that particular RAM stick model.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |