Testing both of them now, Curtis.

Max, could you run these two through CADO (or PM the proper way to do it myself?):
[CODE]# norm 2.436104e-020 alpha -7.803954 e 1.493e-015 rroots 3
skew: 567534086.71
c0: -376103614420928170450425496947906089689878358124495
c1: 230908831907645222124147429766369829700266
c2: 6275884759588926667873913088253117
c3: -13318028314862866742723016
c4: 18451165301092884
c5: 22320144
Y0: -6068679638793086189411803545732110212144
Y1: 39500688790939641031

# norm 2.675402e-020 alpha -8.553644 e 1.593e-015 rroots 3
skew: 715313215.30
c0: -1558062848405312438483027870978047527653011490060415
c1: -185761151115070519611300899870942079387502
c2: 26353663104315594573833027236803953
c3: 322995413784849538032600
c4: -33190654344187596
c5: 22320144
Y0: -6068679638811364632834086638577712163498
Y1: 39500688790939641031[/CODE]

@wombatman
For the last 2 polys
CADO's pick:
[code]
Y0: -6068679638808960167130357221709969339370
Y1: 39500688790939641031
c0: -1472085362277381637401688800900729413934363047079423
c1: 2997790336639737280392043061466727590721298
c2: 25725094131698813758439648214495089
c3: -6931425867968028078651432
c4: -26397352455916236
c5: 22320144
skew: 722871999.33944
# lognorm 67.26, E 58.71, alpha -8.55 (proj -2.53), 3 real roots
# MurphyE(Bf=1.00e+07,Bg=5.00e+06,area=1.00e+16)=1.59272194e-15
[/code]Msieve's confirmation:
[code]
# norm 2.865842e-020 alpha -7.774942 e 1.627e-015 rroots 3
skew: 340123234.89
c0: 571991564598637896614710101239020638310733930025
c1: 167700052788587220002691890175305892882498
c2: 6476627046873628489030255421467789
c3: -13678387964207629928614248
c4: 17897969236909044
c5: 22320144
Y0: 6068679638793281991245086507839408878726
Y1: -39500688790939641031
[/code]

CADO's pick for VBCurtis's polys (same E, same alpha, might sieve differently):
[code]
Y0: -6978726067069333864813640886958939977440
Y1: 55240270001944146943
c0: 8460498597896110004320680623175481689079783021449
c1: 436892324041484604393481553217790456953647
c2: -1479795734643734164968294000358692
c3: -8309136296598625477914295
c4: 5774562363541259
c5: 11099088
skew: 438006794.57101
# alpha -6.47 (proj -1.58), 5 real roots
# MurphyE=1.53325219e-15
[/code]

Trying a degree-6 search for Wombatman's C207. Half an hour of GPU produced four hits worth root-opt, but no polys better than 9e-16. I'm starting from A1=5000, with stage1norm set to 1e26.

I'm interested to see CADO's root-opt for deg 6.

Thanks!

C207 deg 6 polys
 

Attached are a handful of degree-6 polys from msieve with scores between 1.41e-16 and 1.45e-16. They all came from the same hit, A1 = 12408.

I've searched from 5000 to 55000, continuing.

A reminder: deg 6 scores are *not* directly comparable to deg 5. The scores for deg 6 and deg 5 are close enough for test-sieving to be in order.

Oops, forgot attachment.

Thanks a ton! I'll give these a test sieve over night.

In case my meaning wasn't clear, we'd like to know if the sec/rel of the best deg 5 and the best deg 6 are competitive; if they're not, we can rule out one branch of the search- though I think it'll be within 10%.
You might want to wait for Max to have a go at these, though- I hear CADO has made nice progress with deg 6 root-opt, an area Jason admitted wasn't great in msieve.

I've already got it running, so we can see how much difference there is between what MSieve produced and what CADO might output in terms of sieving results.

Best degree 6 poly (by expected time to complete, according to YAFU) was:
[CODE]# norm 5.572602e-15 alpha -10.412567 e 1.458e-15 rroots 4
skew: 19372583.93
c0: 368245649124351475276394529580207588987490339095
c1: -79648633219703333515219159123516394408264
c2: 12617321691323006782072397021881924
c3: -1783238427991767586306863674
c4: -1107519385787745514547
c5: 3647853875938
c6: 12408
Y0: -4955388528018862148437364132620392
Y1: 33855796097504351[/CODE]

with
[CODE]total yield: 283, q=100001029 (4.89875 sec/rel)
test: found 283 relations in a range of 969 special-q
test: estimated total sieving time = 26126 days 18h 12m 58s (with 1 threads)[/CODE]

So the yield is pretty low.

For comparison, the best degree 5 polynomial gave:
[CODE]total yield: 359, q=100001029 (4.56413 sec/rel)
test: found 359 relations in a range of 969 special-q
test: estimated total sieving time = 24338 days 24m 30s (with 1 threads)[/CODE]

So not as far off as I would have expected. I'll try again with further optimized degree 6 ones.