Message ID | 1460011290-28704-1-git-send-email-Shengzhou.Liu@nxp.com |
---|---|
State | Accepted |
Commit | aa7a2226b5a7829915d189d727ee9320dc3a198b |
Delegated to: | York Sun |
Headers | show |
On 04/06/2016 11:50 PM, Shengzhou Liu wrote: > Optimize DDR timing for good margins to support new Transcend > and Apacer DDR4 UDIMM besides current Micron UDIMM. > > Verified 1333MT/s, 1600MT/s, 1866MT/s, 2133MT/s rate with > following UDIMM on LS2080ARDB. > - Micron UDIMM: MTA18ASF1G72AZ-2G1A1Z > - Apacer UDIMM: 78.C1GM4.AF10B > - Transcend UDIMM: TS1GLH72V1H > > Signed-off-by: Shengzhou Liu <Shengzhou.Liu@nxp.com> > --- > v2: verified lower rate, for 1333MT/s no changes are necessary. > > board/freescale/ls2080ardb/ddr.h | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/board/freescale/ls2080ardb/ddr.h b/board/freescale/ls2080ardb/ddr.h > index bda9d4a..b3c6306 100644 > --- a/board/freescale/ls2080ardb/ddr.h > +++ b/board/freescale/ls2080ardb/ddr.h > @@ -29,9 +29,9 @@ static const struct board_specific_parameters udimm0[] = { > * ranks| mhz| GB |adjst| start | ctl2 | ctl3 > */ > {2, 1350, 0, 4, 6, 0x0708090B, 0x0C0D0E09,}, > - {2, 1666, 0, 4, 8, 0x08090B0D, 0x0E10100C,}, > - {2, 1900, 0, 4, 8, 0x090A0C0E, 0x1012120D,}, > - {2, 2300, 0, 4, 9, 0x0A0B0C10, 0x1114140E,}, > + {2, 1666, 0, 5, 9, 0x090A0B0E, 0x0F11110C,}, > + {2, 1900, 0, 6, 0xA, 0x0B0C0E11, 0x1214140F,}, > + {2, 2300, 0, 6, 0xB, 0x0C0D0F12, 0x14161610,}, > {} > }; > > Shengzhou, I see a problem with this patch on one LS2085ARDB board, at 1866MT/s with MT18ASF1G72AZ-2G1A1. Please try it yourself on ls2085rdb-3 (ATX boardfarm). My order of patches are http://patchwork.ozlabs.org/patch/607290/ http://patchwork.ozlabs.org/patch/598135/ http://patchwork.ozlabs.org/patch/598136/ http://patchwork.ozlabs.org/patch/607274/ My test code base is master branch bbca7108db79076d3a9a9c112792d7c4608a665c. York
On 05/05/2016 12:29 PM, York Sun wrote: > On 04/06/2016 11:50 PM, Shengzhou Liu wrote: >> Optimize DDR timing for good margins to support new Transcend >> and Apacer DDR4 UDIMM besides current Micron UDIMM. >> >> Verified 1333MT/s, 1600MT/s, 1866MT/s, 2133MT/s rate with >> following UDIMM on LS2080ARDB. >> - Micron UDIMM: MTA18ASF1G72AZ-2G1A1Z >> - Apacer UDIMM: 78.C1GM4.AF10B >> - Transcend UDIMM: TS1GLH72V1H >> >> Signed-off-by: Shengzhou Liu <Shengzhou.Liu@nxp.com> >> --- >> v2: verified lower rate, for 1333MT/s no changes are necessary. >> >> board/freescale/ls2080ardb/ddr.h | 6 +++--- >> 1 file changed, 3 insertions(+), 3 deletions(-) >> >> diff --git a/board/freescale/ls2080ardb/ddr.h b/board/freescale/ls2080ardb/ddr.h >> index bda9d4a..b3c6306 100644 >> --- a/board/freescale/ls2080ardb/ddr.h >> +++ b/board/freescale/ls2080ardb/ddr.h >> @@ -29,9 +29,9 @@ static const struct board_specific_parameters udimm0[] = { >> * ranks| mhz| GB |adjst| start | ctl2 | ctl3 >> */ >> {2, 1350, 0, 4, 6, 0x0708090B, 0x0C0D0E09,}, >> - {2, 1666, 0, 4, 8, 0x08090B0D, 0x0E10100C,}, >> - {2, 1900, 0, 4, 8, 0x090A0C0E, 0x1012120D,}, >> - {2, 2300, 0, 4, 9, 0x0A0B0C10, 0x1114140E,}, >> + {2, 1666, 0, 5, 9, 0x090A0B0E, 0x0F11110C,}, >> + {2, 1900, 0, 6, 0xA, 0x0B0C0E11, 0x1214140F,}, >> + {2, 2300, 0, 6, 0xB, 0x0C0D0F12, 0x14161610,}, >> {} >> }; >> >> > > Shengzhou, > > I see a problem with this patch on one LS2085ARDB board, at 1866MT/s with > MT18ASF1G72AZ-2G1A1. Please try it yourself on ls2085rdb-3 (ATX boardfarm). My > order of patches are > > http://patchwork.ozlabs.org/patch/607290/ > http://patchwork.ozlabs.org/patch/598135/ > http://patchwork.ozlabs.org/patch/598136/ > http://patchwork.ozlabs.org/patch/607274/ > > My test code base is master branch bbca7108db79076d3a9a9c112792d7c4608a665c. > Shengzhou, I rebased the code and repeated the tests on the same board but seeing no issue now. York
On 05/17/2016 09:25 AM, York Sun wrote: > On 05/05/2016 12:29 PM, York Sun wrote: >> On 04/06/2016 11:50 PM, Shengzhou Liu wrote: >>> Optimize DDR timing for good margins to support new Transcend >>> and Apacer DDR4 UDIMM besides current Micron UDIMM. >>> >>> Verified 1333MT/s, 1600MT/s, 1866MT/s, 2133MT/s rate with >>> following UDIMM on LS2080ARDB. >>> - Micron UDIMM: MTA18ASF1G72AZ-2G1A1Z >>> - Apacer UDIMM: 78.C1GM4.AF10B >>> - Transcend UDIMM: TS1GLH72V1H >>> >>> Signed-off-by: Shengzhou Liu <Shengzhou.Liu@nxp.com> >>> --- >>> v2: verified lower rate, for 1333MT/s no changes are necessary. >>> >>> board/freescale/ls2080ardb/ddr.h | 6 +++--- >>> 1 file changed, 3 insertions(+), 3 deletions(-) >>> >>> diff --git a/board/freescale/ls2080ardb/ddr.h b/board/freescale/ls2080ardb/ddr.h >>> index bda9d4a..b3c6306 100644 >>> --- a/board/freescale/ls2080ardb/ddr.h >>> +++ b/board/freescale/ls2080ardb/ddr.h >>> @@ -29,9 +29,9 @@ static const struct board_specific_parameters udimm0[] = { >>> * ranks| mhz| GB |adjst| start | ctl2 | ctl3 >>> */ >>> {2, 1350, 0, 4, 6, 0x0708090B, 0x0C0D0E09,}, >>> - {2, 1666, 0, 4, 8, 0x08090B0D, 0x0E10100C,}, >>> - {2, 1900, 0, 4, 8, 0x090A0C0E, 0x1012120D,}, >>> - {2, 2300, 0, 4, 9, 0x0A0B0C10, 0x1114140E,}, >>> + {2, 1666, 0, 5, 9, 0x090A0B0E, 0x0F11110C,}, >>> + {2, 1900, 0, 6, 0xA, 0x0B0C0E11, 0x1214140F,}, >>> + {2, 2300, 0, 6, 0xB, 0x0C0D0F12, 0x14161610,}, >>> {} >>> }; >>> >>> >> >> Shengzhou, >> >> I see a problem with this patch on one LS2085ARDB board, at 1866MT/s with >> MT18ASF1G72AZ-2G1A1. Please try it yourself on ls2085rdb-3 (ATX boardfarm). My >> order of patches are >> >> http://patchwork.ozlabs.org/patch/607290/ >> http://patchwork.ozlabs.org/patch/598135/ >> http://patchwork.ozlabs.org/patch/598136/ >> http://patchwork.ozlabs.org/patch/607274/ >> >> My test code base is master branch bbca7108db79076d3a9a9c112792d7c4608a665c. >> > > Shengzhou, > > I rebased the code and repeated the tests on the same board but seeing no issue now. > Shengzhou, I am still seeing it unstable on ls2085rdb-5 (ATX boardfarm) at 1866MT/s. It can boot if I change clk_adjust to 5. Please work with hardware team to confirm. York
On 04/06/2016 11:50 PM, Shengzhou Liu wrote: > Optimize DDR timing for good margins to support new Transcend > and Apacer DDR4 UDIMM besides current Micron UDIMM. > > Verified 1333MT/s, 1600MT/s, 1866MT/s, 2133MT/s rate with > following UDIMM on LS2080ARDB. > - Micron UDIMM: MTA18ASF1G72AZ-2G1A1Z > - Apacer UDIMM: 78.C1GM4.AF10B > - Transcend UDIMM: TS1GLH72V1H > > Signed-off-by: Shengzhou Liu <Shengzhou.Liu@nxp.com> > --- > v2: verified lower rate, for 1333MT/s no changes are necessary. > > board/freescale/ls2080ardb/ddr.h | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) Applied to u-boot-fsl-qoriq master, awaiting upstream. Thanks. York
> -----Original Message----- > From: York Sun [mailto:york.sun@nxp.com] > Sent: Thursday, May 19, 2016 4:28 AM > To: Shengzhou Liu <shengzhou.liu@nxp.com>; u-boot@lists.denx.de > Subject: Re: [U-Boot] [PATCH v2] armv8/ls2080ardb: Update DDR timing to > support more UDIMMs > > > > Shengzhou, > > I am still seeing it unstable on ls2085rdb-5 (ATX boardfarm) at 1866MT/s. It > can boot if I change clk_adjust to 5. Please work with hardware team to > confirm. > > York York On LS1046RDB, the clk_adj is 9, an odd instead of even data, so we have to update it, and there will be more new boards in future. Shengzhou
diff --git a/board/freescale/ls2080ardb/ddr.h b/board/freescale/ls2080ardb/ddr.h index bda9d4a..b3c6306 100644 --- a/board/freescale/ls2080ardb/ddr.h +++ b/board/freescale/ls2080ardb/ddr.h @@ -29,9 +29,9 @@ static const struct board_specific_parameters udimm0[] = { * ranks| mhz| GB |adjst| start | ctl2 | ctl3 */ {2, 1350, 0, 4, 6, 0x0708090B, 0x0C0D0E09,}, - {2, 1666, 0, 4, 8, 0x08090B0D, 0x0E10100C,}, - {2, 1900, 0, 4, 8, 0x090A0C0E, 0x1012120D,}, - {2, 2300, 0, 4, 9, 0x0A0B0C10, 0x1114140E,}, + {2, 1666, 0, 5, 9, 0x090A0B0E, 0x0F11110C,}, + {2, 1900, 0, 6, 0xA, 0x0B0C0E11, 0x1214140F,}, + {2, 2300, 0, 6, 0xB, 0x0C0D0F12, 0x14161610,}, {} };
Optimize DDR timing for good margins to support new Transcend and Apacer DDR4 UDIMM besides current Micron UDIMM. Verified 1333MT/s, 1600MT/s, 1866MT/s, 2133MT/s rate with following UDIMM on LS2080ARDB. - Micron UDIMM: MTA18ASF1G72AZ-2G1A1Z - Apacer UDIMM: 78.C1GM4.AF10B - Transcend UDIMM: TS1GLH72V1H Signed-off-by: Shengzhou Liu <Shengzhou.Liu@nxp.com> --- v2: verified lower rate, for 1333MT/s no changes are necessary. board/freescale/ls2080ardb/ddr.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)