Looking at the first builds where failures started occurring:
https://luci-milo.appspot.com/buildbot/chromium.perf/Mac%20Pro%2010.11%20Perf/881
https://luci-milo.appspot.com/buildbot/chromium.perf/Mac%20Retina%20Perf/583
https://luci-milo.appspot.com/buildbot/chromium.perf/Win%207%20Perf/636
https://luci-milo.appspot.com/buildbot/chromium.perf/Mac%2010.11%20Perf/613
https://luci-milo.appspot.com/buildbot/chromium.perf/Win%208%20Perf/668
https://luci-milo.appspot.com/buildbot/chromium.perf/Win%207%20x64%20Perf/652
https://luci-milo.appspot.com/buildbot/chromium.perf/Win%207%20Nvidia%20GPU%20Perf/661

Came up with the following overlapping CL range:
http://test-results.appspot.com/revision_range?start=466935&end=466937

The following look possibly related (maybe?):
"memory-infra: Start disentangling tracing from memory-infra"
https://codereview.chromium.org/2820433005

"Update V8 to version 6.0.79."
https://codereview.chromium.org/2837163003


While we investigate I'll proceed to disable the story.

From https://luci-logdog.appspot.com/v/?s=chrome%2Fbb%2Fchromium.perf%2FMac_Retina_Perf%2F585%2F%2B%2Frecipes%2Fsteps%2Fsystem_health.memory_desktop_on_Intel_GPU_on_Mac_on_Mac-10.11%2F0%2Fstdout

I see the JS stack trace below
I find a bit surprising that https://codereview.chromium.org/2820433005 could cause a crash within v8.

+bmeurer, +peria could this be one of your changes in v8?

-----

==== JS stack trace =========================================

Security context: 0x3c3649c2b21 <String[24]: https://drive.google.com>
    2: G [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:243] [bytecode=0xf0b45f23ff1 offset=97](this=0x98c4d9d0a41 <Object map = 0x2ea7249bf7c1>#0#,a=0x295c5325b081 <lbb map = 0x2220de55e351>#1#,b=0x2f81f3a82351 <the_hole>,c=0,d=-1,e=0x2f81f3a82201 <null>)
    4: new constructor(aka lbb) [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:2676] [bytecode=0x30a535f4cfb9 offset=27](this=0x295c5325b081 <lbb map = 0x2220de55e351>#1#,a=0x2f81f3a82351 <the_hole>)
    7: L [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:245] [bytecode=0x54496585a11 offset=68](this=0x98c4d9d0a41 <Object map = 0x2ea7249bf7c1>#0#,a=0x295c53250e49 <JSObject>#2#,b=0x9dfed465e1 <JSFunction lbb (sfi = 0x39f3cf0a85d1)>#3#,c=5,d=0x2f81f3a82311 <undefined>)
    8: arguments adaptor frame: 3->4
   10: pO [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:2679] [bytecode=0x30a535f4cda1 offset=27](this=0x1d16c54e01c1 <JSGlobal Object>#4#,a=0x295c5325af61 <Hbb map = 0x2220de55e5b9>#5#)
   12: rO [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:2684] [bytecode=0x30a535f4cb89 offset=17](this=0x1d16c54e01c1 <JSGlobal Object>#4#,a=0x295c5325af61 <Hbb map = 0x2220de55e5b9>#5#)
   14: U [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:2684] [bytecode=0x8d1604830f1 offset=7](this=0x295c5325af61 <Hbb map = 0x2220de55e5b9>#5#,a=15)
   16: Bc [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:2696] [bytecode=0x30a535f49ea1 offset=464](this=0x295c53250a41 <tO map = 0x12cb15f2c9>#6#)
   18: fpa [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:2706] [bytecode=0x7cf0c7f9ad1 offset=239](this=0x3da5b53fb4c1 <vO map = 0x233bc9c90a61>#7#,a=0x295c53250a41 <tO map = 0x12cb15f2c9>#6#,b=0x295c53252009 <Number 1.20425e+12>)
   20: /* anonymous */ [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:7] [bytecode=0x2f5bff8f7d81 offset=49](this=0x3da5b53fb4c1 <vO map = 0x233bc9c90a61>#7#)
   22: dLa [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:1738] [bytecode=0x1d16c54be531 offset=41](this=0x1d16c54e01c1 <JSGlobal Object>#4#,a=0xf80ca9a5ce9 <String[2]: re>,b=0x295c53252171 <JSFunction (sfi = 0x2f5bff8f7a11)>#8#,c=0x3da5b53fb4c1 <vO map = 0x233bc9c90a61>#7#)
   24: c [https://drive.google.com/_/drive_fe/_/js/k=drive_fe.main.en_US.DAxKPHidPdg.O/m=core/am=TO4YKg/rt=j/d=1/rs=AFB8gswQOSH9oxbwlAe3NOTydE5apawiCA:75] [bytecode=0xf0b45f76659 offset=53](this=0x1d16c54e01c1 <JSGlobal Object>#4#)

-----

Looks like we leak the_hole to JavaScript. Could be related to my change. You could revert it speculatively. How do you repro this locally?

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/3721c3608a5aef73b1762e6906257c9d9f8e0d36

commit 3721c3608a5aef73b1762e6906257c9d9f8e0d36
Author: perezju <perezju@chromium.org>
Date: Thu Apr 27 11:53:13 2017

[tools/perf] Disable load:tools:drive system health story

Story shows a high failure rate on many different perf bots.

TBR=nednguyen@google.com
BUG= 715936 

Review-Url: https://codereview.chromium.org/2848573002
Cr-Commit-Position: refs/heads/master@{#467640}

[modify] https://crrev.com/3721c3608a5aef73b1762e6906257c9d9f8e0d36/tools/perf/page_sets/system_health/loading_stories.py

try:

$ tools/perf/run_benchmark.py system_health.memory_desktop --story-filter load:tools:drive --browser [some_browser]

Try "--browser list" to show available browsers, from the configs failing this should repro on a browser for any platform.

 Issue 715561  has been merged into this issue.

Yep, it's my CL (https://codereview.chromium.org/2836913004). Revert in-flight.

The following revision refers to this bug:
  https://chromium.googlesource.com/v8/v8.git/+/53d31d0623bd88ce9a0041a767f36eafd3f1de8b

commit 53d31d0623bd88ce9a0041a767f36eafd3f1de8b
Author: bmeurer <bmeurer@chromium.org>
Date: Fri Apr 28 05:40:46 2017

Revert of [turbofan] General consolidation of element access. (patchset #1 id:1 of https://codereview.chromium.org/2836913004/ )

Reason for revert:
Breaks load:tools:drive story

Original issue's description:
> [turbofan] General consolidation of element access.
>
> Avoid TransitionElementsKind when storing to objects which only differ
> in holeyness of their elements kind. Instead go for polymorphic
> CheckMaps, which can often by optimized and avoid the mutation of the
> array map.
>
> This generalizes the approach https://codereview.chromium.org/2836943003
> which covered only element loads.
>
> R=yangguo@chromium.org
> BUG=v8:5267
>
> Review-Url: https://codereview.chromium.org/2836913004
> Cr-Commit-Position: refs/heads/master@{#44828}
> Committed: https://chromium.googlesource.com/v8/v8/+/ed573cee5c1d1e42158829dc0b92fb697234e121

TBR=yangguo@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=v8:5267, chromium:715936 

Review-Url: https://codereview.chromium.org/2852623002
Cr-Commit-Position: refs/heads/master@{#44958}

[modify] https://crrev.com/53d31d0623bd88ce9a0041a767f36eafd3f1de8b/src/compiler/access-info.cc
[modify] https://crrev.com/53d31d0623bd88ce9a0041a767f36eafd3f1de8b/src/compiler/access-info.h

Issue 715687 has been merged into this issue.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/17cb970f145b4eb3df0c95ced62d38dbf521ca6d

commit 17cb970f145b4eb3df0c95ced62d38dbf521ca6d
Author: primiano <primiano@chromium.org>
Date: Fri Apr 28 11:16:33 2017

Revert of [tools/perf] Disable load:tools:drive system health story (patchset #1 id:1 of https://codereview.chromium.org/2848573002/ )

Reason for re-enabling the story:
The V8 CL that was causing the crash has been reverted in https://crrev.com/2852623002

Original issue's description:
> [tools/perf] Disable load:tools:drive system health story
>
> Story shows a high failure rate on many different perf bots.
>
> TBR=nednguyen@google.com
> BUG= 715936 
>
> Review-Url: https://codereview.chromium.org/2848573002
> Cr-Commit-Position: refs/heads/master@{#467640}
> Committed: https://chromium.googlesource.com/chromium/src/+/3721c3608a5aef73b1762e6906257c9d9f8e0d36

BUG= 715936 

Review-Url: https://codereview.chromium.org/2851703003
Cr-Commit-Position: refs/heads/master@{#467958}

[modify] https://crrev.com/17cb970f145b4eb3df0c95ced62d38dbf521ca6d/tools/perf/page_sets/system_health/loading_stories.py

Issue 718857 has been merged into this issue.