8Gb: 2 Channels x16/x8 GDDR6 SGRAM

Features

GDDR6 SGRAM
MT61K256M32
2 Channels x 256 Meg x 16 I/O, 2 Channels x 512 Meg x 8 I/O

Features • Low power modes

• On‐chip temperature sensor with read‐out
• VDD = V DDQ = 1.35V ±3% and 1.25V ±3% • Auto precharge option for each burst access
• VPP = 1.8V –3%/+6% • Auto refresh mode (32ms, 16k cycles) with per-bank
• Data rate: 12 Gb/s, 14 Gb/s, 16 Gb/s and per-2-bank refresh options
• 2 separate independent channels (x16) • Temperature sensor controlled self refresh rate
• x16/x8 and 2-channel/pseudo channel (PC) mode • Digital tRAS lockout
configurations set at reset • On‐die termination (ODT) for all high‐speed inputs
• Single ended interfaces per channel for command/ • Pseudo open drain (POD135 and POD125) compati-
address (CA) and data ble outputs
• Differential clock input CK_t/CK_c for CA per 2 • ODT and output driver strength auto calibration
channels with external resistor ZQ pin (120Ω)
• One differential clock input WCK_t/WCK_c per • Internal V REF with DFE for data inputs, with input
channel for data (DQ, DBI_n, EDC) receiver characteristics programmable per pin
• Double data rate (DDR) command/address (CK) • Selectable external or internal V REF for CA inputs;
• Quad data rate (QDR) and double data rate (DDR) programmable V REF offsets for internal V REF
data (WCK), depending on operating frequency • Vendor ID for device identification
• 16n prefetch architecture with 256 bits per array • IEEE 1149.1 compliant boundary scan
read or write access • 180-ball BGA package
• 16 internal banks • Lead-free (RoHS-compliant) and halogen-free
• 4 bank groups for tCCDL = 3tCK and 4tCK packaging
• Programmable READ latency • TC = 0°C to +95°C
• Programmable WRITE latency
• Write data mask function via CA bus with single and Options1 Marking
double byte mask granularity • Organization
• Data bus inversion (DBI) and CA bus inversion – 256 Meg × 32 (words × bits) 256M32
(CABI) • FBGA package
• Input/output PLL – 180-ball (12.0mm × 14.0mm) JE
• CA bus training: CA input monitoring via DQ/ • Timing – maximum data rate
DBI_n/EDC signals – 12 Gb/s -12
• WCK2CK clock training with phase information via – 14 Gb/s -14
EDC signals – 16 Gb/s -16
• Data read and write training via read FIFO (depth = • Operating temperature
6) – Commercial (0°C ≤ T C ≤ +95°C) None
• Read/write data transmission integrity secured by • Revision A
cyclic redundancy check using half data rate CRC
• Programmable CRC READ latency Note: 1. Not all options listed can be combined to
• Programmable CRC WRITE latency define an offered product. Use the part
• Programmable EDC hold pattern for CDR catalog search on http://www.micron.com
• RDQS mode on EDC pins for available offerings.

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 1 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
Products and specifications discussed herein are subject to change by Micron without notice.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Features

Figure 1: Part Numbering

MT 61 K 256M32 JE -16 : A
Micron Memory Revision A
Product Family Temperature
61 = GDDR6 SGRAM : = Commercial
Operating Voltage Data Rate
K = 1.35V -12 = 12 Gb/s
-14 = 14 Gb/s
Configuration -16 = 16 Gb/s
256M32 = 256 Meg x 32
Package
JE = 180-ball FBGA, 12.0mm x 14.0mm

FBGA Part Marking Decoder

Due to space limitations, FBGA-packaged components have an abbreviated part marking that is different from the
part number. For a quick conversion of an FBGA code, see the FBGA Part Marking Decoder on Micron’s web site:
http://www.micron.com.

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 2 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Ball Assignments and Descriptions

Ball Assignments and Descriptions

Figure 2: 180-Ball FBGA (Top View)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

A VDD VSS DQ1_A VSS VPP V PP VSS DQ9_A VSS VDD A

B VSS DQ3_A DQ2_A DQ0_A VDDQ V DDQ DQ8_A DQ10_A DQ11_A VSS B

C VDDQ EDC0_A VSS VDDQ VSS V SS VDDQ V SS EDC1_A VDDQ C

VSS DBI0_n_A VSS WCK_t WCK_c VSS VSS

D NC NC DBI1_n_A D
_A _A

E VDD Q DQ5_A DQ4_A VSS VDD V DD VSS DQ12_A DQ13_A VDDQ E

F VSS DQ6_A VSS VDDQ TMS TDI V DDQ VSS DQ14_A VSS F

G VSS DQ7_A VSS CA2_A NC CKE_n_A CA1_A VSS DQ15_A VSS G

H VDDQ VDD CA0_A VSS CA4_A CA5_A VSS CA3_A VDD VDDQ H

J RESET _n VDDQ CA9_A CA8_A CABI_n_A CK_t CA7_A CA6_A VDDQ ZQ_A J

K VREFC VDDQ CA9_B CA8_B CABI_n_B C K _c CA7_B CA6_B VDDQ ZQ_B K

L VDDQ VDD CA0_B VSS CA4_B CA5_B VSS CA3_B VDD VDDQ L

M VSS DQ7_B VSS CA2_B NC CKE_n_B CA1_B VSS DQ15_B VSS M

N VSS DQ6_B VSS VDDQ TCK TDO VDDQ VSS DQ14_B VSS N

P VDDQ DQ5_B DQ4_B VSS VDD V DD VSS DQ13_B VDDQ P

DQ12_B

R VSS DBI0_n_B VSS NC NC WCK_c WCK_t VSS DBI1_n_B VSS R

_B _B

T VDDQ EDC0_B VSS VDDQ VSS VSS VDDQ VSS EDC1_B VDDQ T

U VSS DQ3_B DQ2_B DQ0_B VDDQ VDDQ DQ8_B DQ10_B DQ11_B VSS U

V VDD VSS DQ1_B VSS VPP V PP VSS DQ9_B VSS VDD V

Command/
Data Other signal Supply Ground
Address

Note: 1. Channel A byte 1 and channel B byte 0 are disabled when the device is configured to x8
mode.

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 3 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Ball Assignments and Descriptions

Table 1: 180-Ball FBGA Ball Descriptions

Symbol Type Description

CK_t, Input Clock: CK_t and CK_c are differential clock inputs. CK_t and CK_c do not have chan-
CK_c nel indicators as one clock is shared between both channel A and channel B on a de-
vice. Command address (CA) inputs are latched on the rising and falling edge of CK.
All latencies are referenced to CK.
WCK_t, Input Write clock: WCK_t and WCK_c are differential clocks used for write data capture
WCK_c and read data output. WCK_t/WCK_c are associated with DQ[15:0], DBI[1:0]_n, and
EDC[1:0].
CKE_n Input Clock enable: CKE_n LOW activates and CKE_n HIGH deactivates the internal clock,
device input buffers, and output drivers excluding RESET_n, TDI, TDO, TMS, and TCK.
Taking CKE_n HIGH provides PRECHARGE POWER-DOWN and SELF REFRESH opera-
tions (all banks idle), or ACTIVE POWER-DOWN (row ACTIVE in any bank). CKE_n
must be maintained LOW throughout read and write accesses.
CA[9:0] Input Command address (CA): The CA inputs receive packetized DDR command, address
or other information, for example, the op-code for the MRS command. See Com-
mand Truth Table for details.
CABI_n Input Command address bus inversion
DQ[15:0] I/O Data input/output: Bidirectional 16-bit data bus.
DBI[1:0]_n I/O Data bus inversion: DBI0_n is associated with DQ[7:0], DBI1_n is associated with
DQ[15:8].
EDC[1:0] Output Error detection code: The calculated CRC data is transmitted on these signals. In
addition these signals drive a "hold" pattern when idle. EDC0 is associated with
DQ[7:0], EDC1 is associated with DQ[15:8].
VDDQ Supply I/O power supply: Isolated on the die for improved noise immunity.
VDD Supply Power supply
VSS Supply Ground
VPP Supply Pump voltage
VREFC Supply Reference voltage for CA, CABI_n, and CKE_n signals
ZQ Reference External reference for auto calibration
TDI Input JTAG test data input
TDO Output JTAG test data output
TMS Input JTAG test mode select
TCK Input JTAG test clock
RESET_n Input Reset: RESET_n low asynchronously initiates a full chip reset. With RESET_n LOW all
ODTs are disabled. A full chip reset may be performed at any time by pulling RE-
SET_n LOW.
NC – No connect

Note: 1. Index "_A" or "_B" represents the channel indicator "A" and "B" of the device. Signal
names including the channel indicator are used whenever more than one channel is ref-
erenced, for example, with the ball assignment. The channel indicator is omitted when-
ever features and functions common to both channels are described.

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 4 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Package Dimensions

Package Dimensions

Figure 3: 180-Ball FBGA (JE)

0.12

Seating plane

A 0.1 A
0.6 CTR
nonconductive
overmold

180X Ø0.47
Dimensions apply
to solder balls post- Ball A1 ID Ball A1 ID
reflow on Ø0.42 SMD (covered by SR)
ball pads. 14 13 12 11 10 5 4 3 2 1
A
B
C
D
E
F
G
14 ±0.1 H
J
12.75 CTR K
L
M
N
P
R
T
U
0.75 TYP V

1.1 ±0.1
0.75 TYP

9.75 CTR 0.34 ±0.05

12 ±0.1

Notes: 1. Package dimension specification is compliant to JC11 MO328 variation P14.0x12.0-

GJ-180A.
2. All dimensions are in millimeters.
3. Solder ball material: SAC-Q (92.5% Sn, 4% Ag, 3% Bi, 0.5% Cu).

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 5 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Functional Description

Functional Description
The GDDR6 SGRAM is a high-speed dynamic random-access memory designed for ap-
plications requiring high bandwidth. It is internally configured as 16‐bank memory and
contains 8,589,934,592 bits.
The GDDR6 SGRAM’s high-speed interface is optimized for point-to-point connections
to a host controller. On-die termination (ODT) is provided for all high-speed interface
signals to eliminate the need for termination resistors in the system.
GDDR6 uses a 16n-prefetch architecture and a DDR or QDR interface to achieve high-
speed operation. The device’s architecture consists of two 16-bit-wide fully independ-
ent channels.
Read and write accesses to GDDR6 are burst oriented; accesses start at a selected loca-
tion and consist of a total of 16 data words. Accesses begin with the registration of an
ACTIVATE command, which is then followed by a READ, WRITE (WOM), or masked
WRITE (WDM, WSM) command. The row and bank address to be accessed is registered
coincident with the ACTIVATE command. The address bits registered coincident with
the READ, WRITE, or masked WRITE command are used to select the bank and the
starting column location for the burst access.

Clocking
GDDR6 operates from a differential clock CK_t and CK_c. CK is common to both chan-
nels. Command and address (CA) are registered at every rising and falling CK edge.
There are both single-cycle and multi-cycle commands. See Command Truth Table for
details.
GDDR6 uses a free running differential forwarded clock (WCK_t/WCK_c) with both in-
put and output data registered and driven respectively at both edges of the forwarded
WCK.
GDDR6 supports DDR and QDR operating modes for WCK frequency which differ in
the DQ/DBI_n pin to WCK clock frequency ratio. The figure below illustrates the differ-
ence between both modes.
This GDDR6 SGRAM device is designed with a WCK/word granularity which is equiva-
lent to one WCK per channel. The DRAM info bits for WCK granularity, WCK frequency,
and internal WCK can be read by the host during the initialization process to determine
the WCK architecture for the device.

Table 2: Example Clock and Interface Signal Frequency Relationship

Pin DDR WCK QDR WCK Unit

CK_t, CK_c 1.5 1.5 GHz
CA 3.0 3.0 Gb/s/pin
WCK_t, WCK_c 6.0 3.0 GHz
DQ, DBI_n 12.0 12.0 Gb/s/pin
EDC 6.0 6.0 Gb/s/pin

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 6 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Functional Description

Figure 4: Clocking and Interface Relationship

CK_c
f (for example, 1.5 GHz)
CK_t

CA 2f (for example, 3 Gb/s)

QDR WCK 2f (for example, 3 GHz)

DDR WCK 4f (for example, 6 GHz)

DQ, DBI_n 8f (for example, 12 Gb/s)

EDC 4f (for example, 6 Gb/s)

Note: 1. The figure shows the relationship between the data rate of the buses and the clocks; it
is not a timing diagram.

Figure 5: Block Diagram of an Example Clock System

Controller GDDR6 SGRAM

ADD/CMD centered with CK_t/CK_c ADD/CMD sampled by CK_t/CK_c as DDR

to internal
CMD/ADD DQ DQ state machine
CA
(3 Gb/s)

WCK2CK
CK_t, CK_c Alignment
(1.5 GHz)

DQ to EDC pin

PLL /2
Osc. PLL Internal WCK
Data Tx /Rx WCK_t, WCK_c /4
3.0 GHz
(3 GHz or
6 GHz)

READ
data QD QD
clock DRAM
phase early/
ctrl late Core
DATA
WRITE
data DQ (12 Gb/s) DQ

clock
phase
ctrl

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 7 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Functional Description

Addressing
GDDR6 addressing is defined for a single channel with devices having two channels per
device.

Table 3: Addressing

8Gb Density
Parameter x16 Mode x8 Mode
Number of channels 2
Memory density (per channel) 4Gb
Memory prefetch (per channel) 256b 128b
Bank address (per channel) BA[3:0]
Row address (per channel) R[13:0]
Column address (per channel) C[5:0] C[6:0]
Page size (per channel) 2KB
Refresh 16k/32ms

Notes: 1. The column address notation for GDDR6 does not include the lower four address bits as
the burst order is always fixed for READ and WRITE.
2. Page size = 2^COLBITS × (Prefetch_Size/8) where COLBITS is the number of column ad-
dress bits.

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 8 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Functional Description

Operations
Command Truth Table
GDDR6 uses a packetized DDR command/address bus that encodes all commands and
addresses on a 10-bit CA bus as outlined in the table below.

Figure 6: Command Truth Table

CKE_n
CK
Operation Symbol CA9 CA8 CA7 CA6 CA5 CA4 CA3 CA2 CA1 CA0 Notes
Edge
n-1 n

NO OPERATION NOP (1) R L L H H V V V V V V V V 1, 10

F H H V V V V V V V V
NOP (2) R L L H H V V V V V V V V
F H L V V V V V V V V

NOP (3) R L L H L V V V V V V V V
F H H V V V V V V V V
MODE REGISTER SET MRS R L L H L M3 M2 M1 M0 OP3 OP2 OP1 OP0 1, 2, 3
F H L OP11 OP10 OP9 OP8 OP7 OP6 OP5 OP4
ACTIVATE ACT R L L L V BA3 BA2 BA1 BA0 R3 R2 R1 R0 1, 2, 4
F R13 R12 R11 R10 R9 R8 R7 R6 R5 R4
READ RD R L L H H BA3 BA2 BA1 BA0 C3 C2 C1 C0 1, 2, 5,
L H L L V L CE C6 C5 C4 6
F
READ with RDA R L L H H BA3 BA2 BA1 BA0 C3 C2 C1 C0 1, 2, 5,
AUTO PRECHARGE 6
F L H L L V H CE C6 C5 C4
LOAD FIFO LDFF R L L H H B3 B2 B1 B0 D3 D2 D1 D0 1, 2, 8
F L H H L D9 D8 D7 D6 D5 D4
READ TRAINING RDTR R L L H H V V V V V V V V 1, 2, 6
F L H H H V L CE V V V
WRITE WOM R L L H H BA3 BA2 BA1 BA0 C3 C2 C1 C0 1, 2, 5,
6
F L L L L V L CE C6 C5 C4
WRITE with WOMA R L L H H BA3 BA2 BA1 BA0 C3 C2 C1 C0 1, 2, 5,
AUTO PRECHARGE 6
F L L L L V H CE C6 C5 C4
WRITE SINGLE WSM R L L H H BA3 BA2 BA1 BA0 C3 C2 C1 C0 1, 2, 5,
BYTE MASK 6
F L L L H V L CE C6 C5 C4

H H Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0

R
BST7 BST6 BST5 BST4 BST3 BST2 BST1 BST0

H H Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0

F BST15 BST14 BST13 BST12 BST11 BST10 BST9 BST8

H H Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1

R BST7 BST6 BST5 BST4 BST3 BST2 BST1 BST0

H H Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1

F
BST15 BST14 BST13 BST12 BST11 BST10 BST9 BST8

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 9 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Functional Description

Figure 7: Command Truth Table (Continued)

CKE_n
CK
Operation Symbol CA9 CA8 CA7 CA6 CA5 CA4 CA3 CA2 CA1 CA0 Notes
Edge
n-1 n

WRITE SINGLE WSMA R L L H H BA3 BA2 BA1 BA0 C3 C2 C1 C0 1, 2, 5,

BYTE MASK with 6
F L L L H V H CE C6 C5 C4
AUTO PRECHARGE
H H Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0
R
BST7 BST6 BST5 BST4 BST3 BST2 BST1 BST0

H H Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0 Byte 0

F
BST15 BST14 BST13 BST12 BST11 BST10 BST9 BST8

H H Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1

R
BST7 BST6 BST5 BST4 BST3 BST2 BST1 BST0

H H Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1 Byte 1

F BST15 BST14 BST13 BST12 BST11 BST10 BST9 BST8

WRITE DOUBLE WDM R L L H H BA3 BA2 BA1 BA0 C3 C2 C1 C0 1, 2, 5,

BYTE MASK 6
F L L H L V L CE C6 C5 C4
R H H BST7 BST6 BST5 BST4 BST3 BST2 BST1 BST0
F H H BST15 BST14 BST13 BST12 BST11 BST10 BST9 BST8
WRITE DOUBLE WDMA R L L H H BA3 BA2 BA1 BA0 C3 C2 C1 C0 1, 2, 5,
BYTE MASK with 6
F L L H L V H CE C6 C5 C4
AUTO PRECHARGE
R H H BST7 BST6 BST5 BST4 BST3 BST2 BST1 BST0
F H H BST15 BST14 BST13 BST12 BST11 BST10 BST9 BST8
WRITE TRAINING WRTR R L L H H V V V V V V V V 1, 2, 6
F L L H H V L CE V V V
PRECHARGE PREpb R L L H L BA3 BA2 BA1 BA0 V V V V 1, 2, 9
F L L V V V L V V V V
PRECHARGE ALL PREab R L L H L V V V V V V V V 1, 2
F L L V V V H V V V V
PER-BANK REFRESH REFpb/ R L L H L BA3 BA2 BA1 BA0 V V V V 1, 2, 7,
REFp2b 9
F L H V V V L V V V V
REFRESH REFab R L L H L V V V V V V V V 1, 2, 7
F L H V V V H V V V V
POWER-DOWN ENTRY PDE R L H H H V V V V V V V V 1, 2
F H H V V V V V V V V
POWER-DOWN EXIT PDX R H L H H V V V V V V V V 1, 2
F H H V V V V V V V V
SELF REFRESH ENTRY SRE R L H H L V V V V V V V V 1, 2, 7
F L H V V V V V V V V
SELF REFRESH EXIT SRX R H L H H V V V V V V V V 1, 2
F H H V V V V V V V V
COMMAND/ADDRESS CAT R L H V V V V V V V V V V 1, 2
TRAINING CAPTURE
F V V V V V V V V V V

Notes: 1. H = Logic HIGH level; L = Logic LOW level; V = Valid signal (H or L, but not floating). R, F
= Rising, Falling CK clock edge.
CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 10 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Functional Description

2. Values shown for CA[9:0] are logical values; the physical values are inverted when com-
mand/address bus inversion (CABI) is enabled and CABI_n = L.
3. M[3:0] provide the mode register address (MRA), OP[11:0] the opcode to be loaded.
4. BA[3:0] provide the bank address, R[13:0] provide the row address.
5. B[3:0] provide the bank address, C[6:0] provide the column address; no sub-word ad-
dressing within a burst of 16. BST[15:0] provide the write data mask for each burst posi-
tion with WDM(A) and WSM(A) commands.
6. CE (channel enable) is intended for PC mode. The command is active when CE = H.
When CE = L the data access is suppressed.
7. The command is REFRESH or PER-BANK REFRESH/PER-2-BANK REFRESH when CKE_n(n) =
L and SELF REFRESH ENTRY when CKE_n(n) = H.
8. B[3:0] select the burst position, and D[9:0] provide the data.
9. BA[3:0] provide the bank address.
10. All three encodings perform the same NOP. NOP (2) and NOP (3) encodings are only al-
lowed during CA Training.

Clamshell (x8) Mode Enable

A GDDR6 SGRAM-based memory system is typically divided into several channels.
GDDR6 has been optimized for a 16-bit-wide channel. A channel can be comprised of a
single device operated in x16 mode, or two devices each operated in x8 mode. For x8
mode the devices are typically assembled on opposite sides of the PCB in what is refer-
red as a clamshell layout.
Whether in x16 mode or x8 mode the device will operate with a point-to-point connec-
tion on the high-speed data signals. The disabled signals in x8 mode should all be in a
High-Z state, non-terminating.
The x8 mode is detected at power-up on EDC1_A and EDC0_B. For x8 mode these sig-
nals are tied to V SS; they are part of the bytes that are disabled in this mode and there-
fore not needed for EDC functionality. For x16 mode these signals are active and always
terminated to V DDQ in the system or by the controller.
The configuration is set with RESET_n going HIGH. Once the configuration has been
set, it cannot be changed during normal operation. Typically, the configuration is fixed
in the system. Details of the x8 mode detection are depicted in Figure 8. A comparison
of x16 mode and x8 mode systems is shown in Figure 9.

Table 4: Clamshell (x8) Mode Enable

Mode EDC0_A EDC1_A EDC0_B EDC1_B

x8 VDDQ VSS (on board) VSS (on board) VDDQ
x16 VDDQ (terminated by the system or controller)

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 11 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Functional Description

Figure 8: Enabling Clamshell (x8) Mode

Controller VDDQ GDDR6

EDC data from in x8 mode
Enable other DRAM non-mirrored
Termination

EDC1_A
EDC0_B EDC Data
RX EDC TX
EDC1_A EN
EDC0_B x8

RX D 0 = x8
VSS
RESET_n RESET_n
RESET_n

Controller VDDQ GDDR6

EDC data from in x8 mode
Enable other DRAM mirrored
Termination

EDC1_A
EDC0_B EDC Data
RX EDC TX
EDC0_A EN
EDC1_B x8
0 = x8
VSS RX D

RESET_n RESET_n
RESET_n

Controller VDDQ GDDR6

in x16 mode
Enable
Termination

EDC1_A
EDC0_B EDC Data
RX EDC TX
EDC1_A EN
EDC0_B x16

RX D 1 = x16
VSS
RESET_n RESET_n
RESET_n

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 12 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Functional Description

Figure 9: System View for x16 and x8 Modes

GDDR6 GDDR6 GDDR6

Host Host x8
x16 x8

DQ[7:0]_X, DBI0_n_X Byte 0_A DQ[7:0]_X, DBI0_n_X Byte 0_A

EDC0_X EDC0_A EDC0_X EDC0_A EDC0_B
Channel X

Channel X
WCK_t_X, WCK_c_X WCK_A WCK0_t_X, WCK0_c_X WCK_A
WCK1_t_X, WCK1_c_X WCK_B
DQ[15:8]_X, DBI1_n_X Byte 1_A
DQ[15:8]_X, DBI1_n_X Byte 1_B
EDC1_X EDC1_A
EDC1_X EDC1_B
EDC1_A
ADD/ ADD/ ADD/
CA[9:0]_X, CABI_n_X CMD CA[9:0]_X, CABI_n_X CMD CMD
CKE_n_X (Ch A) CKE_n_X (Ch A) (Ch B)

DQ[15:8]_Y, DBI1_n_Y Byte 1_B DQ[15:8]_Y, DBI1_n_Y Byte 1_B

EDC1_Y EDC1_B EDC1_Y EDC1_B EDC1_A

WCK_t_Y, WCK_c_Y WCK_B WCK1_t_Y, WCK1_c_Y WCK_B
Channel Y

Channel Y
WCK0_t_Y, WCK0_c_Y WCK_A
DQ[7:0]_Y, DBI0_n_Y Byte 0_B
EDC0_Y EDC0_B DQ[7:0]_Y, DBI0_n_Y Byte 0_A
EDC0_Y EDC0_A
EDC0_B
ADD/ ADD/ ADD/
CA[9:0]_Y, CABI_n_Y CMD CA[9:0]_Y, CABI_n_Y
CMD CMD
CKE_n_Y (Ch B) CKE_n_Y (Ch B) (Ch A)
CK_t, CK_c CK_t, CK_c
RESET_n RESET_n

Figure 10 clarifies the use of x8 mode and how the bytes are enabled/disabled to give
the controller the view of the same bytes that a controller sees with a single x16 device.
For a 16-bit channel using two devices in a clamshell design, byte 0 comes from channel
A from the top device and byte 1 comes from channel B from the bottom device and will
look equivalent at the controller to a x16 mode.

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 13 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Functional Description

Figure 10: Byte Orientation in Clamshell Topology

Ch A Ch B Ch A Ch A
Byte Byte Byte Byte
0 1 0 1
+ =
Ch B Ch A Ch B Ch B
Byte Byte Byte Byte
1 0 0 1

x8 top
x8

x8
x8 bottom
Legend:
Data
ADD/CMD
CK, WCK

Pseudo-Channel Mode
GDDR6 has been optimized for a 32B access across a 16-bit channel by providing a
unique CA bus to each 16-bit-wide channel. For applications requiring fewer CA pins,
GDDR6 includes support for a pseudo-channel (PC) mode where CA[9:4], CKE_n, and
CABI_n on each channel are connected to a common bus, while CA[3:0] of each chan-
nel are connected to a separate bus. The command truth table is organized such that in
PC mode the same command is decoded in both pseudo-channels, but READ and
WRITE commands support a unique column address to each pseudo-channel. In PC
mode, CKE_n and CABI_n are also shared across pseudo-channels.
In PC mode, the only difference in the DRAM is that termination on CA[9:4], CKE_n,
and CABI_n can be configured differently from CA[3:0]. PC mode can be selected during
initialization by driving CA6 = LOW on both channels when RESET_n is driven HIGH.

Figure 11: CA Pins in Pseudo-Channel Mode

Pseudo-Channel Mode Controller GDDR6

CA[3:0]_A CA[3:0]_A
CA[9:4]_A, CABI_n_A, CKE_n_A
CA[9:4], CABI_n, CKE_n
CA[9:4]_B, CABI_n_B, CKE_n_B
CA[3:0]_B CA[3:0]_B

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 14 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Operating Conditions

Operating Conditions
Absolute Maximum Ratings
Stresses greater than those listed may cause permanent damage to the device. This is a
stress rating only, and functional operation of the device at these or any other condi-
tions above those indicated in the operational sections of this specification is not im-
plied. Exposure to absolute maximum rating conditions for extended periods may ad-
versely affect reliability.

Table 5: Absolute Maximum Ratings

Symbol Parameter Min Max Unit Notes

VDD Voltage on VDD pin relative to VSS –0.3 2.0 V 1
VDDQ Voltage on VDDQ pin relative to VSS –0.3 2.0 V 1
VPP Voltage on VPP pin relative to VSS –0.3 2.3 V 2
VIN/VOUT Voltage on any pins relative to VSS –0.3 2.0 V
TSTG Storage temperature –55 +125 °C

Notes: 1. VDD and VDDQ must be within 300mV of each other at all times the device is powered‐
up.
2. VPP must be equal or greater than VDD and VDDQ at all times the device is powered‐up.

DC and AC Operating Conditions

The interface of GDDR6 with 1.35V V DDQ will follow the POD135 Standard (JESD8-21),
Class D; The interface with 1.25V V DDQ will follow the POD125 Standard (JESD8-30),
Class A. All AC and DC values are referenced to the ball.

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 15 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN
CCMTD-1412786195-10191

Table 6: DC Operating Conditions

POD135 POD125
Symbol Parameter Min Typ Max Min Typ Max Unit Notes
VDD Device supply voltage 1.3095 1.35 1.3905 1.2125 1.25 1.2875 V 1, 2
VDDQ Output supply voltage 1.3095 1.35 1.3905 1.2125 1.25 1.2875 V 1, 2
VPP Pump voltage 1.746 1.8 1.908 1.746 1.8 1.908 V 2
VREFD Reference voltage for DQ and DBI_n 0.69 × VDDQ – 0.71 × VDDQ 0.69 × VDDQ − 0.71 × VDDQ V 3, 4
VREFD2 0.49 × VDDQ – 0.51 × VDDQ 0.49 × VDDQ − 0.51 × VDDQ V 3, 4, 5
VREFC Reference voltage for CA 0.69 × VDDQ – 0.71 × VDDQ 0.69 × VDDQ − 0.71 × VDDQ V 3, 6
VREFC2 0.49 × VDDQ – 0.51 × VDDQ 0.49 × VDDQ − 0.51 × VDDQ V 3, 6, 7
VIHA(DC) DC input logic HIGH voltage with VREFC for CA VREFC + – – VREFC + − − V
0.135 0.125
VILA(DC) DC input logic LOW voltage with VREFC for CA – – VREFC ‐ 0.135 − − VREFC - 0.125 V
VIHA2(DC) DC input logic HIGH voltage with VREFC2 for CA VREFC2 + – – VREFC2 + − − V
0.27 0.25
VILA2(DC) DC input logic LOW voltage with VREFC2 for CA – – VREFC2 ‐ 0.27 − − VREFC2 - 0.25 V
16

VIHD(DC) DC input logic HIGH voltage with VREFD for DQ and VREFD + 0.09 – – VREFD + − − V
DBI_n 0.085

8Gb: 2 Channels x16/x8 GDDR6 SGRAM

VILD(DC) DC input logic LOW voltage with VREFD for DQ and – – VREFD ‐ 0.09 − − VREFD - 0.085 V
DBI_n
Micron Technology, Inc. reserves the right to change products or specifications without notice.

VIHD2(DC) DC input logic HIGH voltage with VREFD2 for DQ and VREFD2 + – – VREFD2 + − − V
DBI_n 0.27 0.25
VILD2(DC) DC input logic LOW voltage with VREFD2 for DQ and – – VREFD2 ‐ 0.27 − − VREFD2 - 0.25 V
DBI_n
VIHR RESET_n and boundary scan input logic HIGH volt- 0.8 × VDDQ – – 0.8 × VDDQ − − V 8
age; EDC and CA input logic HIGH voltage for x16/x8

Operating Conditions
mode, PC vs. 2-channel mode, CK and CA ODT select
© 2016 Micron Technology, Inc. All rights reserved.

at reset
VILR RESET_n and boundary scan input logic LOW voltage; – – 0.2 × VDDQ − − 0.2 × VDDQ V 8
EDC and CA input logic LOW voltage for x16/x8
mode, PC vs. 2-channel mode, CK and CA ODT select
at reset
VIN Single ended clock input voltage level: CK_t, CK_c, –0.30 – VDDQ + 0.30 –0.30 − VDDQ + 0.30 V 14
WCK_t, WCK_c
VMP(DC) CK_t, CK_c clock input midpoint voltage VREFC - 0.10 – VREFC + 0.10 VREFC ‐ 0.10 − VREFC + 0.10 V 9, 12
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN
CCMTD-1412786195-10191

Table 6: DC Operating Conditions (Continued)

POD135 POD125
Symbol Parameter Min Typ Max Min Typ Max Unit Notes
VIDCK(DC) CK_t, CK_c clock input differential voltage 0.198 – – 0.18 − − V 10, 12
VIDWCK(DC) WCK_t, WCK_c clock input differential voltage 0.18 – – 0.165 − − V 11, 13
IL Input leakage current (any input 0V ≤ VIN ≤ VDDQ; all –5 – 5 −5 − 5 µA
other signals not under test = 0V)
IOZ Output leakage current (outputs are disabled; 0V ≤ –5 – 5 −5 − 5 µA
VOUT ≤ VDDQ)
VOL(DC) Output logic low voltage – – 0.56 – – 0.52 V
ZQ External resistor value 115 120 125 115 120 125 Ω

Notes: 1. GDDR6 SGRAM devices are designed to tolerate PCB designs with separate VDD and VDDQ power regulators.
2. DC bandwidth is limited to 20 MHz.
3. AC noise in the system is estimated at 50mV peak-to-peak for the purpose of DRAM design.
4. The reference voltage source and control for DQ and DBI_n pins are determined by half VREFD, and VREFD level
mode register bits.
5. Programmable VREFD levels are not supported with VREFD2.
17

6. The reference voltage source (external or internal) is determined at power‐up; the reference voltage level is deter-
mined by half VREFC and the VREFC offset mode register bit.

8Gb: 2 Channels x16/x8 GDDR6 SGRAM

7. Programmable VREFC offsets are not supported with VREFC2.
8. VIHR and VILR apply to boundary scan input pins TDI, TMS, and TCK. VIHR and VILR apply to EDC and CA inputs at
Micron Technology, Inc. reserves the right to change products or specifications without notice.

reset when latching default device configurations. VIHR and VILR also apply to CA, CABI_n, CKE_n, CK, DQ, DBI_n,
EDC, and WCK inputs when boundary scan mode is active and input data are latched in the capture-DR TAP con-
troller state.
9. This provides a minimum of 0.845V to a maximum of 1.045V with POD135, and a minimum of 0.775V to a maxi-
mum of 0.975V with POD125, and is normally 70% of VDDQ. DRAM timings relative to CK cannot be guaranteed if
these limits are exceeded.
10. VIDCK is the magnitude of the difference between the input level in CK_t and the input level on CK_c. The input

Operating Conditions
reference level for signals other than CK_t and CK_c is VREFC.
© 2016 Micron Technology, Inc. All rights reserved.

11. VIDWCK is the magnitude of the difference between the input level in WCK_t and the input level on WCK_c. The
input reference level for signals other than WCK_t and WCK_c is either VREFC, VREFC2, VREFD, or VREFD2.
12. The CK_t and CK_c input reference level (for timing referenced to CK_t and CK_c) is the point at which CK_t and
CK_c cross. Refer to the applicable timings in the AC Timings table.
13. The WCK_t and WCK_c input reference level (for timing referenced to WCK_t and WCK_c) is the point at which
WCK_t and WCK_c cross. Refer to the applicable timings in the AC Timings table.
14. Use VIHR and VILR when boundary scan mode is active and input data are latched in the capture-DR TAP controller
state.
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN
CCMTD-1412786195-10191

Table 7: AC Operating Conditions (For Design Only9)

POD135 POD125
Symbol Parameter Min Typ Max Min Typ Max Unit Notes
VIHA(AC) AC input logic HIGH voltage with VREFC for CA VREFC + 0.18 – – VREFC + − − V
0.165
VILA(AC) AC input logic LOW voltage with VREFC for CA – – VREFC ‐ 0.18 − − VREFC - 0.165 V
VIHA2(AC) AC input logic HIGH voltage with VREFC2 for CA VREFC2 + – – VREFC + − − V
0.36 0.333
VILA2(AC) AC input logic LOW voltage with VREFC2 for CA – – VREFC2 ‐ 0.36 − − VREFC - 0.333 V
VIHD(AC) AC input logic HIGH voltage with VREFD for DQ, VREFD + – – VREFD + − − V
DBI_n 0.135 0.125
VILD(AC) AC input logic LOW voltage with VREFD for DQ, DBI_n – – VREFD ‐ 0.135 − − VREFD - 0.125 V
VIHD2(AC) AC input logic HIGH voltage with VREFD2 for DQ, VREFD2 + – – VREFD2 + − − V
DBI_n 0.36 0.333
VILD2(AC) AC input logic LOW voltage with VREFD2 for DQ, – – VREFD2 ‐ 0.36 − − VREFD2 - V
DBI_n 0.333
VIDCK(AC) CK_t, CK_c clock differential voltage 0.36 – – 0.333 − − V 1, 3, 5
18

VIDWCK(AC) WCK_t, WCK_c clock input differential voltage 0.27 – – 0.25 − − V 1, 4, 6

VIXCK(AC) CK_t, CK_c clock input crossing point voltage VREFC - 0.108 – VREFC + VREFC ‐ 0.10 − VREFC + 0.10 V 1, 2, 5

8Gb: 2 Channels x16/x8 GDDR6 SGRAM

0.108
Micron Technology, Inc. reserves the right to change products or specifications without notice.

VIXWCK(AC) WCK_t, WCK_c clock input crossing point voltage VREFD - 0.09 – VREFD + 0.09 VREFC ‐ 0.09 − VREFC + 0.09 V 1, 2, 6,
7

Notes: 1. For AC operations, all DC clock requirements must be satisfied as well.

2. The value of VIXCK and VIXWCK is expected to equal 70% VDDQ for the transmitting device and must track variations
in the DC level of the same.
3. VIDCK is the magnitude of the difference between the input level on CK_t and the input level on CK_c. The input

Operating Conditions
reference level for signals other than CK_t and CK_c is VREFC.
4. VIDWCK is the magnitude of the difference between the input level on WCK_t and the input level on WCK_c. The
© 2016 Micron Technology, Inc. All rights reserved.

input reference level for signals other than WCK_t and WCK_c is either VREFC, VREFC2, VREFD, or VREFD2.
5. The CK_t and CK_c input reference level (for timing referenced to CK_t and CK_c) is the point at which CK_t and
CK_c cross. Refer to the applicable timings in the AC Timings table.
6. The WCK_t and WCK_c input reference level (for timing referenced to WCK_t and WCK_c) is the point at which
WCK_t and WCK_c cross. Refer to the applicable timings in the AC Timings table.
7. VREFD is either VREFD or VREFD2.
8. Figure 13 illustrates the exact relationship between (CK_t - CK_c) or (WCK_t - WCK_c) and VID(AC), VID(DC).
9. The AC operating conditions are for DRAM design only and are valid on the silicon at the input of the receiver.
They are not intended to be measured.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Operating Conditions

Figure 12: Voltage Waveform

VDDQ

VOH
System noise margin (power/ground,
crosstalk, ISI, attenuation)
VIH(AC)

VIH(DC)

VREF + AC noise
VREF + DC error
VREF - DC error
VREF - AC noise

VIL(DC)

VIL(AC)
VIN(AC) provides margin
between VOL(MAX) and
VIL(MAX)
VOL,max
Input

Output

Note: 1. VREF, VIH, and VIL refer to whichever VREFxx (VREFD, VREFD2, VREFC, or VREFC2) is being used.

Figure 13: Clock Waveform

Maximum clock level

CK_t

VID(AC)
VMP(DC) VIX(AC) V
ID(DC)

CK_c

Minimum clock level

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 19 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.
 8Gb: 2 Channels x16/x8 GDDR6 SGRAM
Operating Conditions

8000 S. Federal Way, P.O. Box 6, Boise, ID 83707-0006, Tel: 208-368-4000

www.micron.com/products/support Sales inquiries: 800-932-4992
Micron and the Micron logo are trademarks of Micron Technology, Inc.
All other trademarks are the property of their respective owners.
This data sheet contains minimum and maximum limits specified over the power supply and temperature range set forth herein.
Although considered final, these specifications are subject to change, as further product development and data characterization some-
times occur.

CCMTD-1412786195-10191
gddr6_sgram_8gb_brief.pdf - Rev. F 8/18 EN 20 Micron Technology, Inc. reserves the right to change products or specifications without notice.
© 2016 Micron Technology, Inc. All rights reserved.