Oh please, don't even try that line of argument. I am a software developer for a very large automotive company (I do lots of CPU intensive simulations and HIL testing), and I can tell you with absolute confidence that if my company entered into a business relationship with AMD, no amount of input from the engineers will allow them to get an Intel system.
It's true, I went from having a nice and sweet 24-node AMD cluster for running my sims at a university (where I had full reign over my procurement decisions) to having to build a replacement cluster based on Intel-only (ended up being a P4 system, thankfully Northwood so it wasn't excessively D:, but still...) solely because the new employer was DELL exclusive AND Intel exclusive on top of that.
Even if DELL offered AMD kit, my employer's IT dept told me straight up they would not support the cluster if I bought non-Intel gear. I resented the fact I was hamstrung of choices, but I consoled myself with the fact that I was getting my own $100k cluster dedicated solely to my sims and that kind of capex pampering is hard to get to fussed up about.
For the same performance I could have had a $50k AMD cluster...but it wasn't my money and the people who were authorizing me to spend their money did have a preference, a pricier preference. Bad for shareholders, same fun for IDC, and in the end the company made money off the results from the cluster anyways regardless that it was $50k over-priced.
That's business. What are you going to do?
Exactly, and that is the crux of the problem. Obviously it would be better for AMD to cut the cord with BAPCo than be associated to a benchmark with an intel overlord, which doesn't evaluate their hardware to it's capabilities. If it compels only a handful of procurment decision makers to look elsewhere to evaluate their needs, then it's mission accomplished. They would have lost those sales anyway since the results of the evaluating software does no justice to AMD's new features. It seems that many analysts agree that Sysmark has run it's course and it's relevancy has past. Anyway, I guess this horse is dead.
I had not considered the residual benefits, that they stood to nab a few more contracts that would have otherwise been assured losses had they stood by silently without protest against Sysmark.
It is definitely the case that these benchmarks seem to come and go in cycles.
SPEC and LINPACK are probably some of the longest lasting, but many other benches seem to live and die within the span of a decade. Sysmark lasted more than a decade but it wasn't a decade without questionable practices and fallout thereof.