F5F Stay Refreshed Hardware Desktop Good 8Gb DDR4 RAM suitable for around 4400 MHz is available.

Good 8Gb DDR4 RAM suitable for around 4400 MHz is available.

Good 8Gb DDR4 RAM suitable for around 4400 MHz is available.

Pages (3): Previous 1 2 3 Next
D
DanielEmpire
Posting Freak
781
12-15-2017, 05:42 AM
#11
I've noticed DDR4 5000 screenshots from the 2666 range, but I don't think JEDEC categories are used for this.
D
DanielEmpire
12-15-2017, 05:42 AM #11

I've noticed DDR4 5000 screenshots from the 2666 range, but I don't think JEDEC categories are used for this.

T
TorJules
Member
184
12-15-2017, 05:42 AM
#12
It seems 'd die' is quite frequent, so if I can't locate 'c9bjz' or 'c9bhs rev e,' maybe I should try 'd die.' I'm also curious about direct comparisons between 'd die' and 'rev e'—just want to know which performs better, especially since I'm leaning toward 'rev e.'
T
TorJules
12-15-2017, 05:42 AM #12

It seems 'd die' is quite frequent, so if I can't locate 'c9bjz' or 'c9bhs rev e,' maybe I should try 'd die.' I'm also curious about direct comparisons between 'd die' and 'rev e'—just want to know which performs better, especially since I'm leaning toward 'rev e.'

C
Cptnbacon
Junior Member
31
12-15-2017, 05:42 AM
#13
The main reference I recall is the one mentioned in a previous post. He compares two different ICs using an Aida analysis. Rev. E shows marginally better latency, while D die has slightly higher bandwidth due to a higher clock speed. Keep in mind he's using Coffee Lake, which means Rev. E may not have strong support there, so results could differ.
C
Cptnbacon
12-15-2017, 05:42 AM #13

The main reference I recall is the one mentioned in a previous post. He compares two different ICs using an Aida analysis. Rev. E shows marginally better latency, while D die has slightly higher bandwidth due to a higher clock speed. Keep in mind he's using Coffee Lake, which means Rev. E may not have strong support there, so results could differ.

W
WD_Trashster
Senior Member
454
12-15-2017, 05:42 AM
#14
D dies stops scaling around 1.6 volts. The B350 gaming K4 lacks VTTDR adjustment, so it will likely be limited to that voltage.
W
WD_Trashster
12-15-2017, 05:42 AM #14

D dies stops scaling around 1.6 volts. The B350 gaming K4 lacks VTTDR adjustment, so it will likely be limited to that voltage.

S
Sofi41505
Member
149
12-15-2017, 05:42 AM
#15
Stops near 1.8V roughly. Most likely the only factors affected by voltage are frequency and CAS latency. If you're not focusing on reducing CAS latency, setting it above 1.5–1.6V probably isn't necessary.
S
Sofi41505
12-15-2017, 05:42 AM #15

Stops near 1.8V roughly. Most likely the only factors affected by voltage are frequency and CAS latency. If you're not focusing on reducing CAS latency, setting it above 1.5–1.6V probably isn't necessary.

J
Jayden32805
Member
212
12-15-2017, 05:42 AM
#16
Trfc behavior depends on the voltage levels; your DDR2 and DDR3 I/O appear affected by it. Other components may also respond to voltage changes, so adjusting frequency and subs can help stabilize performance.
J
Jayden32805
12-15-2017, 05:42 AM #16

Trfc behavior depends on the voltage levels; your DDR2 and DDR3 I/O appear affected by it. Other components may also respond to voltage changes, so adjusting frequency and subs can help stabilize performance.

D
Dormin15
Member
110
12-15-2017, 05:42 AM
#17
Not on D die. Results consistently reach around 300ns, matching Rev. E. Similar for the primaries without tCL and most other subtimings. Some DDR4 like B die works, but newer models generally don't, unless specified. It's not worth trying to push it above 1.5V.
D
Dormin15
12-15-2017, 05:42 AM #17

Not on D die. Results consistently reach around 300ns, matching Rev. E. Similar for the primaries without tCL and most other subtimings. Some DDR4 like B die works, but newer models generally don't, unless specified. It's not worth trying to push it above 1.5V.

O
oHits
Member
176
12-15-2017, 05:42 AM
#18
I notice a good number of double-sided 16GB bare PCBs with revision E and D dies. I'm wondering why these tend to be different from non-bare PCBs, since single-sided ones usually have better quality. I'm also a bit worried about double-sided boards—maybe some only have one slot per channel or a daisy-chained layout that doesn't work well with four sticks.
O
oHits
12-15-2017, 05:42 AM #18

I notice a good number of double-sided 16GB bare PCBs with revision E and D dies. I'm wondering why these tend to be different from non-bare PCBs, since single-sided ones usually have better quality. I'm also a bit worried about double-sided boards—maybe some only have one slot per channel or a daisy-chained layout that doesn't work well with four sticks.

I
II_RIPTIDE_II
Member
167
12-15-2017, 05:42 AM
#19
Varies with the components, OEM Hynix parts usually show looser timing compared to G.Skill with Hynix, especially on DDR5 where I've spent most of my time recently. Good PCBs don’t usually make a huge difference, but they do matter. Regarding double-sided performance, it should perform similarly to standard G.Skill boards if the chips can handle it. The only real-world examples I’ve seen are ECC server RAM, some DDR3 (primarily Samsung 2Gb D die), and a few single-sided DDR4 sticks—so I haven’t had direct experience with those.
I
II_RIPTIDE_II
12-15-2017, 05:42 AM #19

Varies with the components, OEM Hynix parts usually show looser timing compared to G.Skill with Hynix, especially on DDR5 where I've spent most of my time recently. Good PCBs don’t usually make a huge difference, but they do matter. Regarding double-sided performance, it should perform similarly to standard G.Skill boards if the chips can handle it. The only real-world examples I’ve seen are ECC server RAM, some DDR3 (primarily Samsung 2Gb D die), and a few single-sided DDR4 sticks—so I haven’t had direct experience with those.

C
51
12-15-2017, 05:42 AM
#20
Patriot Viper Steel 4400-19-19-19 offers a good value for money on Amazon.
C
clement_246871
12-15-2017, 05:42 AM #20

Patriot Viper Steel 4400-19-19-19 offers a good value for money on Amazon.

Pages (3): Previous 1 2 3 Next