[COMPLETED] Extended Downtime Notification - 21.01.2021

Thank you! It was very fast!

Tranquility is back online and accepting connections now! Please let us know if you experience any issues with accessing New Eden :slight_smile: Fly safe!

7 Likes

Apparently, from an article I read, CCP feels that they need faster processors, rather than update their code to use more processor cores and threadsā€¦

I was surprised about thatā€¦ code can be written to use more cores and threads as they become available, but they didnā€™t allow for it.

I donā€™t get itā€¦

1 Like

I think they have so much crappy code from the earlier days that itā€™s easier to upgrade the hardware than untangle the a ball of snot.

No CCP have in their own way said that will not be fixed.

Hands off my essential backend.

1 Like

No, translated from CCP language the technician chair in the server room had to be replaced, unfortunately servers were in the way umd had to be turned off briefly. :nerd_face:

1 Like

I love you guys. more posts here than on the MER thread

1 Like

Well, it can be salvaged, and there is new tech that can be employed.

The question is, will CCP leverage it? The Imperium and PAPI are currently locked into a kind of limbo. Anyone that logs into those accounts gets killed, on either side. Stalemate. And CCP is trying to figure out how to salvage it.

I donā€™t have any details on their server structure, but, clearly, they opted for faster processors rather than allowing for multi-core growth.

Now I get that as a business, going for low-cost hardware is sensible, but, I for one would have gone for parallel processing, like the Beowulf cluster concept: add a new node and the cluster rebalances the load automatically. Upgrade the node with more cores, the node rebalances to include the new cores and threads.

I donā€™t know if theyā€™ve considered this or not, or that they have a hardware/software plan, but it seems sensible to me to bet on both sides of the development curve. Intel and AMD are better informed than most on how to get more MIPS out of a processor, and I would have paid the money to consult with either to plan for the future. Right now, Intel and AMD is aiming for ā€œon chip clusteringā€ to improve speed since theyā€™ve nearly reached the limit on the nanoscale.

Of course, thereā€™s a company that has taken things to extremes: clusters on a wafer. That may be the next step. An entire cluster on a wafer, rather than on a network, complete with memory storage. Iā€™m talking about hundreds of multi-core, multi-threaded processors on a single production wafer in a single box, with the benefits of on chip communication speeds. Cheap? No, but you might be able to have battles the size that are beyond expectations.

Of course, the question is, what are they willing to consider? Or, do they have the original code they started with?

The one thing I learned, many moons agoā€¦ code is only as good as the hardware it runs on, and how portable it is.

1 Like

This topic was automatically closed after 30 hours. New replies are no longer allowed.