AMD 3D V-Cache RAM Disk Delivers Over 182 GB/s and 175 GB/s Read and Write Speeds

[HD64G]

Just a metric of speed without practical use. 3D cache is used where needed and helps much there. Other than that, a fun test this.

 

[THU31]

And the sad thing is, if you put an Unreal Engine game on there, you'd still get shader compilation and traversal stutters.

 

[TheGuruStud]

Damn... When I read the title, I was shocked and thought, "What product is AMD making now? A new type of SSD? What a bully wanting to succeed in all the areas that intel failed." But it's just a gimmick.

Theoretically, they could make an existing LGA chock full of these and stacked on each other as high as thermals allow (dual socket system). It would be so fast that your dick would fly off.

 

[chrcoluk]

This is very nice but I expect the advantage of cache is latency rather than the bandwidth.

Need a benchmark testing the latency compared to a normal RAM disk and I expect it will be blowing the RAM out of the water.

 

[AnotherReader]

so, the speed of these L3 cache is just fast as RAM
because right now, with quad channel 6000MHz DDR5, you can already reach 192GB/s

The L3 is vastly faster than DRAM. The figure in the CrystalDiskMark test is the bandwidth available to a single core from the L3. The bandwidth scales with the number of cores. A 7950X gets over 1800 GB/s from L3 with all cores reading from the L3. On the other hand, the 7950X only gets 72.85 GB/s from a 128-bit bus to DDR5-6000. Compare the throughput from a hypothetical 256-bit bus to that L3 result and you'll see why L3 is used. I haven't even touched upon the gap in latency between L3 and DRAM, but that one is even more insurmountable.

 

[A Computer Guy]

The L3 is vastly faster than DRAM. The figure in the CrystalDiskMark test is the bandwidth available to a single core from the L3. The bandwidth scales with the number of cores. A 7950X gets over 1800 GB/s from L3 with all cores reading from the L3. On the other hand, the 7950X only gets 72.85 GB/s from a 128-bit bus to DDR5-6000. Compare the throughput from a hypothetical 256-bit bus to that L3 result and you'll see why L3 is used. I haven't even touched upon the gap in latency between L3 and DRAM, but that one is even more insurmountable.

View attachment 324114

3950x wow. Still a great CPU.