[BuildBot] Uplift GPU RT version for Linux CI Process

[bb-sycl]

Uplift GPU RT version for Linux to 20.47.18513

[bader]

Tests are failing.
We should either wait for a driver update fixing the regression or apply a workaround I suggested here: #2807 (comment)

[dm-vodopyanov]

Please check if PR #2851 fixes the uplifting after being merged.

[bader]

Please check if PR #2851 fixes the uplifting after being merged.

Please, add your patch to this PR, CI system can validate the fix.

[bader]

It works. @dm-vodopyanov, I suggest closing #2851 and commit both patches from this PR if #2851 pre-commit pass. What do you think?

[dm-vodopyanov]

@bader sounds good to me, closed #2851
Upd. Sorry, forgot to wait for Jenkins/Summary status. Can it be run for this PR?

[bader]

Can it be run for this PR?

You can still use #2851 results for loader version update validation.
But it must be done for this PR for a different reason - we run these tests any driver update.

[bader]

/summary:run

[yanfeng3721]

Regression detected by post commit CTS test, the mismatch information is as below:
[2020-12-03T04:56:53.545Z] 48: --- math_builtin_float_double
[2020-12-03T04:56:59.299Z] 48: . value: 0.8986363237370140755189140691072680056095123291015625 [3fecc1a0f6ae5732], reference: 0.89863632373701440858582145665423013269901275634765625 [3fecc1a0f6ae5735]
[2020-12-03T04:56:59.299Z] 48: . line: 152
[2020-12-03T04:56:59.299Z] 48: . tests case: 4600071. Correctness check failed.
[2020-12-03T04:57:38.741Z] 48: - fail

[yanfeng3721]

Hi @bader , the same test can pass with 20.46.18421, so could it possible to reopen #2807 for uplift?

Regression detected by post commit CTS test, the mismatch information is as below:
[2020-12-03T04:56:53.545Z] 48: --- math_builtin_float_double
[2020-12-03T04:56:59.299Z] 48: . value: 0.8986363237370140755189140691072680056095123291015625 [3fecc1a0f6ae5732], reference: 0.89863632373701440858582145665423013269901275634765625 [3fecc1a0f6ae5735]
[2020-12-03T04:56:59.299Z] 48: . line: 152
[2020-12-03T04:56:59.299Z] 48: . tests case: 4600071. Correctness check failed.
[2020-12-03T04:57:38.741Z] 48: - fail

[bader]

@yanfeng3721, it's quite possible a CTS bug, not DPC++.
I'm okay to merge #2807, while this issue is being investigated.

[bader]

To fix SYCL-CTS failures.