Recent content by i386

Yes, apple is selling products that consists of hard- and software. Decoupling the software from the hardware would hurt the brand.

This is the max. ambient temperature. The asic/cpu can get up to 80-90°C before sensors and alarms goes off. https://www.broadcom.com/products/storage/raid-controllers/megaraid-sas-9361-8i

Why? Satadoms are great, you don't lose high performance pcie lanes for a boot device, don't take up two hotswap bays and still benefit from the random io performance of a ssd.

I had a similar experience with conect-x3 nics, I think it's because the nic firmware doesn't know the "new" specifications for 25/50/100GBE transceivers in dacs/aocs/optical modules.

+1 for the toolless adapters MCP-220-00118-0B from supermicro:

Which os? Did you check for partitions/volumes on the device?

About 836 vs 846: I have a 836 and a 846 and if I get another chassis it will be a 846 again because it's "stoage density" is better (24 devices/4u vs 16 devices/3u). Another point for the 846: there are more (sas2 AND sas3) expanders available on ebay than for the 836. Edit: "A" (sas multilane...

Firmware v3.21 + driver v106.190.4.1062 + maxview v3.06.23821 can't handle any storage configuration in windows or pcie changes (adding/removing or moving pcie devices to another slot); the logical drives will disappear and only show up after saving the configuration in the bios. I can...

Not sure if you consider a single socket comparable: I am running windows 10 pro for workstations on a x10sra with a xeon 1630v4, 64GByte ram and a 1080ti (got it for free :P) and play occasionally fps (doom/doom eternal, csgo and the wolfenstein games). So far no differences to a consumer system :D

2.58M x 512 Bytes = ~1.23 GByte/s, that's really impressive for a non nvdimm blockdevice. I'm curios what numbers are possible with random writes. (I was kinda disappointed of first gen optane nvdimms https://arxiv.org/pdf/1903.05714.pdf & First Optane Performance tests show benefits and limits...

It depends how the backplanes are wired :D There are two backplanes, one has 3 multilane sas ports, the other has 2 multilane sas ports. In the 36 bay version I would connect the second backplane with a single cable to the third multilane sas port on the first backplane and then route the two...

Testing would require enough storage devices that could saturate a single link. I would just look at the controllers link status & enclosure info: (This is from my Adaptec/Microchip raid controller)

It's often described/explained as "dual link" : two multilane sas ports on the hba/raid controller connect to two multilane sas ports on the backplane increasing the total bandwidth between expander and hba/raid controller.

That's only true for empty hdds when data is written to the outer sectors. That's why all the datasheets say "up to" or "max" xxx MByte/s. The more data is stored, the slower the drive becomes. Example seagate exos 16TByte: empty ~245MByte/s, ~80% full ~100MByte/s, ~95% full <70MByte/s Edit...

The CS-1 outperformed #82 of the top500 supercomputer by a factor of 200:eek: https://www.cerebras.net/beyond-ai-for-wafer-scale-compute-setting-records-in-computational-fluid-dynamics/ Riesenprozessor schlägt Supercomputer (article is in german!)