Platform Storage Face-Off: AMD Upsets Intel

[Admin]

We put AMD and Intel systems head-to-head to find out which platform provides the top performance with storage devices.

Platform Storage Face-Off: AMD Upsets Intel : Read more

 

[Giroro]

Is it really necessary to keep calling Intel and AMD "The blue team" and "the red team"?
The actual company names are shorter, easier to type, and probably more accessible to color blind readers...
Maybe I'm just exhausted with every aspect of society being cut into US vs THEM teams

 

[paul prochnow]

We put AMD and Intel systems head-to-head to find out which platform provides the top performance with storage devices.

I know I like what I got: R7 2700 [a Ryzen] and an SSD M.2----NVMe ADATASX8200PNP------------ [ which recently became BARGAIN priced] BUT I did not know that I had a winner. As an untutored End Use, a real Mr. GIU, it is great to see someone would got to such great lengths and great detail to help the consumer--THANK YOU AGAIN TOM'S


Platform Storage Face-Off: AMD Upsets Intel : Read more

 

[paul prochnow]

Is it really necessary to keep calling Intel and AMD "The blue team" and "the red team"?
The actual company names are shorter, easier to type, and probably more accessible to color blind readers...
Maybe I'm just exhausted with every aspect of society being cut into US vs THEM teams

Look---It is a matter of someone typing that all out . It saves one keystroke. Ask a corporate manager knows, we all need to know, what a few keystrokes more or less really matter.....I think in State Mangement alone, for instance, one less "e" in EmployEE as in employe saved a million bux in printing costs. Believe it of NOT. I could not, unfortunately THEY could PROVE it. WOW!

 

[USAFRet]

What I get out of that is that the magical NVMe drives (970 EVO) aren't much 'faster' than the typical SATA III drives, such as the 860 EVO.

 

[supremelaw]

raw bandwidth of DDR4-3200 x 8 = 25,600 MB/second

1x PCIe 4.0 lane @ 16 GHz / 8.125 (jumbo frame) x 16 lanes = 31,507 MB/second
(e.g. ASRock Ultra Quad M.2 x16 card in "4x4" mode, enhanced for PCIe 4.0)

31,507 / 8 ~= DDR4-3,938

Thus, at PCIe 4.0, an upgraded ASRock Ultra Quad M.2 x16 card
will have a raw bandwidth equivalent to DDR4-3,938.

The numbers above show that the PCIe 4.0 bus is roughly comparable
the DRAM bus using mid- to high-end DDR4 DRAM.

Now, do the same calculations using the 32G clock planned for PCIe 5.0:

1x PCIe 5.0 lane @ 32 GHz / 8.125 (jumbo frame) x 16 lanes = 63,015 MB/second

63,015 / 8 ~= DDR4-7,876 (i.e. easily surpassing DDR4 maxima at present)

Despite all of the bad press that Intel has been getting recently,
I really must credit them for developing Optane technologies,
which now promise to mass-market a non-volatile memory solution
that operates at speeds roughly comparable to mid-range DRAM.

Putting Optane on the DRAM bus is a very significant development,
for which AMD has no competition.

Keep your eyes open for Samsung's entry into the PCIe 4.0 competition!
You can bet they are planning such entry(s) with excellent foresight.

 

[TerryLaze]

Putting Optane on the DRAM bus is a very significant development,
for which AMD has no competition.

And you are not just getting a fast disk,the more important part about optane ram is that the CPU is a part of the equation and can access data on that ram directly ,skipping a big part of windows I/O.

 

[bit_user]

I disagree with giving Intel a "win" for boot time. Since the article focuses on I/O performance, I wouldn't include the POST time in the contest, although it was informative to measure and report it.

Also, if you understand a bit about why the side-channel attack mitigations impact I/O performance, then it should come as no surprise that small, semi-random I/O's take the worst hit. The synthetics show this, pretty clearly. For most real-world use cases, such as app loading, loading game levels, CAD models, and OS booting, it's this random, small I/O that counts. Only the disk imaging and large file-copying remain largely unaffected.

BTW, the BIOS settings (such as scanning for disks and other devices) have a lot to do with POST time. It's often possible to tweak your BIOS and get that number down. I'd also expect to see different POST times for different brands that use the same chipset.

 

[pct8750]

Is that 9700k new stepping?
plz compare P0 with R0.

 

[Giroro]

Look---It is a matter of someone typing that all out . It saves one keystroke.

I mean... It's a 9 keystroke difference. Which makes "the red team" 4 times the length of AMD.
It's just an odd trend. It reduces the clarity of the language, is harder to type, and nobody is out there referring to companies by color-team in real life so it makes the article flow more like those comparative "OUR PRODUCT VS BRAND X" style commercials than natural speech.

I just don't get it, man.

 

[bit_user]

It's just an odd trend. It reduces the clarity of the language, is harder to type, and nobody is out there referring to companies by color-team in real life ...

I just don't get it, man.

I think it's mostly a journalistic/writing practice of trying to mix up the terminology to make it sound less tedious and formulaic.

Plus, if you see people talk about red vs. blue, then such articles help explain what they mean. And yes, I do sometimes see forum posters refer to the companies (or their fans) by color.